Ci2010Jul011724424296629 USGS Risk Reduction USGSC 200

Implications for
Earthquake Risk Reduction

in the United States from the
Kocaeli, Turkey, Earthquake
of August 17, 1999
U.S. Department of the Interior
U.S. Geological Survey
U.S. Geological Survey Circular 1193
U
.
S
.

G
e
o
l
o
g
i
c
a
l

S
u
r
v
e
y
I
m
p
l
i
c
a
t
i
o
n
s

f
o
r

E
a
r
t
h
q
u
a
k
e

R
i
s
k

R
e
d
u
c
t
i
o
n

i
n

t
h
e

U
.
S
.

f
r
o
m

t
h
e

K
o
c
a
e
l
i
,

T
u
r
k
e
y
,

E
a
r
t
h
q
u
a
k
e

U
.
S
.

G
e
o
l
o
g
i
c
a
l

S
u
r
v
e
y

C
i
r
c
u
l
a
r

1
1
9
3

B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
F
O
L
D
F
O
L
D
Cover: Damage in Korfez, Turkey, following the
August 17 Kocaeli earthquake.
Photograph by Charles Mueller
Cover design by Carol A. Quesenberry
B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
F
O
L
D
F
O
L
D
Field investigations were coordinated with the U.S. Army Corps of Engineers
and National Institute of Standards and Technology
Implications for
Earthquake Risk Reduction
in the United States from the
Kocaeli, Turkey, Earthquake
of August 17, 1999
U.S. Geological Survey Circular 1193
By U.S. Geological Survey
B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
F
O
L
D
F
O
L
D
B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
F
O
L
D
F
O
L
D
Bruce Babbitt, Secretary
Charles G. Groat, Director
Free on application to:
Information Services
Box 25286, Federal Center
Denver, CO 80225
Any use of trade, product, or firm names in this publication is for descriptive purposes only
and does not imply endorsement by the U.S. Government
Contributors
Thomas L. Holzer, Scientific Editor, U.S. Geological Survey
Aykut A. Barka, Istanbul Technical University, Turkey
David Carver, U.S. Geological Survey
Mehmet elebi, U.S. Geological Survey
Edward Cranswick, U.S. Geological Survey
Timothy Dawson, San Diego State University and Southern California Earthquake Center
James H. Dieterich, U.S. Geological Survey
William L. Ellsworth, U.S. Geological Survey
Thomas Fumal, U.S. Geological Survey
John L. Gross, National Institute of Standards and Technology
Robert Langridge, U.S. Geological Survey
William R. Lettis, William Lettis & Associates
Mark Meremonte, U.S. Geological Survey
Charles Mueller, U.S. Geological Survey
Richard S. Olsen, Engineer Research and Development Center, U.S. Army Corps of Engineers
Oguz Ozel, Kandilli Observatory and Earthquake Research Institute, Bogazici University, Turkey
Tom Parsons, U.S. Geological Survey
Long T. Phan, National Institute of Standards and Technology
Thomas Rockwell, San Diego State University and Southern California Earthquake Center
Erdal Safak, U.S. Geological Survey
Ross S. Stein, U.S. Geological Survey
Heidi Stenner, U.S. Geological Survey
Shinji Toda, Earthquake Research Institute, University of Tokyo, Japan
Selcuk Toprak, U.S. Geological Survey
United States Government Printing Office: 2000
B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
F
O
L
D
F
O
L
D
Contents
Summary
Introduction
The Earthquake
Geologic Setting and Earthquake History
Earthquake Forecasts
Surface Faulting
Ground Shaking
Main-Shock Observations
Implications of Aftershock Observations
Liquefaction
Performance of the Built Environment
Buildings
Trans European Motorway (TEM)
Earth Dams
Implications for the United States
Acknowledgments
References
1
3
5
9
11
13
21
21
31
41
47
51
37
41
48
57
55
61
63
EPICENTER
Izmit Dzce
BLACK SEA
SEA OF MARMARA

Black Sea
Greece
AFRICAN PLATE
ARABIAN
PLATE

EURASIAN PLATE

Mediterranean Sea
Turkey
Crete

1999
1967
1944 1943
1939
1942
1951
1957
San Francisco
Los Angeles
Izmit
TURKEY
CALI FORNI A

1

Summary

The moment magnitude (M

W

) 7.4 Kocaeli, Turkey, earthquake struck
the Kocaeli province of northwestern Turkey on Tuesday, August 17, 1999, at
3:02 a.m. local time. The cause of the earthquake was the sudden breakage, or
rupture, of the Earths crust along a western branch of the 1,500-km-long
North Anatolian fault system. The total length of the fault rupture was about
110 km. The region hit by the earthquake is the industrial heartland and the
most densely populated section of Turkey. According to ofcial Turkish gov-
ernment estimates, the earthquake caused 17,127 deaths and 43,953 injuries,
and displaced more than 250,000 people. Approximately 121 tent cities were
required for emergency housing. Approximately 214,000 residential units and
30,500 business units were lightly to heavily damaged.
Most of the collapsed and damaged buildings were of reinforced con-
crete construction. Their poor performance was due primarily to the poor
quality of construction and failure to enforce the local building code. Peak
ground accelerations recorded close to the fault were below average for a M

W

7.4 event. The earthquakes strong velocity pulse and long duration, however,
were important factors in the collapse of and damage to residences.
Immediately following the earthquake, the USGS was invited by the
Kandilli Observatory and Earthquake Research Institute of Bogazici Univer-
sity in Istanbul, Turkey, to assist in its post-earthquake investigations. Team-
ing up with representatives from the National Institute of Standards and
Technology and the U.S. Army Corps of Engineers, the reconnaissance teams
documented the following implications for earthquake risk reduction in the
United States.

Large-scale human disaster:

Of all natural disasters, large-magnitude
earthquakes in urban areas are almost unique in their ability to cause
large-scale human suffering and property loss. The Kocaeli earthquake
damaged many homes and displaced more than a quarter of a million
people. Earthquake loss models predict that tragedies of comparable
scope are possible here in the United States.

Building collapse hazard:

More than 20,000 reinforced concrete build-
ings with little or no ductility are estimated to have collapsed in the
Kocaeli earthquake. Much of the building stock in the United States
was constructed before the importance of ductility, the ability to deform
without catastrophic loss of strength, was fully appreciated. In the
Eastern United States most reinforced concrete structures, including
both buildings and bridges, have little or no ductility. This leaves the
United States vulnerable to the catastrophic collapse of large numbers
of buildings and bridges when they are shaken by the oscillatory motion
of large earthquakes.

Earthquake monitoring systems:

A delayed reaction by government
emergency response ofcials in Turkey to the earthquake disaster may
have been due in part to the inability of the local seismic monitoring
network to determine a reliable seismic magnitude and shaking intensi-
ties for the Kocaeli earthquake. Near real-time warnings using modern
digital technology are now available and can identify precisely both the
size and location of a large earthquake within minutes of its occurrence
as well as the distribution of shaking intensity.

1

Text.fm Page 1 Monday, January 24, 2000 8:32 AM

2

Earthquake forecasting:

The North Anatolian fault system is
a close analog of the San Andreas fault system in California.
The Kocaeli earthquake was the tenth in a systematic west-
ward progression of earthquakes along 1000 km of the
North Anatolian fault since 1939. Understanding how these
earthquakes interacted may be useful for improving proba-
bilistic earthquake forecasts in the United States. The
November 12, 1999, M

W

7.1 earthquake, which was near
Dzce at the eastern end of the August surface rupture,
occurred in an area where the Kocaeli earthquake is esti-
mated to have increased the earthquake potential by increas-
ing stress on the fault.

Permanent ground deformation hazards:

Although earth-
quake shaking typically causes more than 90 percent of the
loss associated with earthquakes directly, earthquake-
induced permanent deformation or shifting of the ground
can be more devastating to structures locally. Three types of
permanent ground deformation were noteworthy in the
Kocaeli earthquake: liquefaction, surface faulting, and den-
sication of dry approach lls on the Trans European Motor-
way. Liquefaction, the transformation of solid material into
liquidlike material, caused many buildings to sink into the
ground. The sinking of buildings suggests a failure mecha-
nism that may be very important to river levee systems in the
United States, which are typically built in areas susceptible
to liquefaction. This hazard compounds the already formi-
dable challenge of dealing with the extensive potential for
static failures of levee systems in the United States. Lique-
faction of silts in Turkey has important implications for
earth dams founded on alluvial foundations around the
world. Although earth dams in Turkey performed well, the
bearing capacity failures of buildings indicates low-strength
silt foundations beneath earth dams can liquefy and cause
the embankment to sink or rotate into the foundation. Sur-
face faulting damaged many buildings as well as the modern
Trans European Motorway, which was built to the same
standard as American highway systems. Densication of dry
approach lls was extensive along the Trans European
Motorway and required its closing to resurface much of the
roadway surface. Permanent ground deformation is not con-
sidered by the current code for highway and bridge design in
the United States.


3

Introduction

3

Ductility:

The property of a material to
deform without catastrophic loss of
strength.

Liquefaction:

The process by which
sandy wet earth materials become
uidlike when shaken by earthquakes.

The Kocaeli, Turkey, earthquake struck near the city of Izmit on August
17, 1999. According to ofcial government estimates (as of October 19,
1999), it killed 17,127, and injured more than 43,953 people. Estimates of
property losses (as of September 14, 1999) according to the World Bank range
from $3 to $6.5 billion, which is equivalent to 1.5 to 3.3 percent of the Gross
National Product of Turkey. It was the most devastating earthquake to strike
Turkey since the 1939 Erzincan earthquake killed 30,000 people. Immedi-
ately following the Kocaeli earthquake, the Kandilli Observatory and Earth-
quake Research Institute of Bogazici University in Istanbul, Turkey, invited
the U.S. Geological Survey (USGS) to assist in its post-earthquake investiga-
tion. In addition to assisting Turkish colleagues, the invitation provided an
opportunity to learn lessons for reducing the risk of earthquakes in the United
States. The USGS team coordinated its efforts in Turkey with representatives
from the Engineer Research and Development Center of the U.S. Army Corps
of Engineers and the National Institute of Standards and Technology.
It was immediately recognized that the Kocaeli earthquake, which
occurred on the North Anatolian fault system, might provide useful insight
into the seismic implications of the geologic record preserved along the San
Andreas fault system in California. Thorough mapping of the surface faulting
in Turkey, thus, had high priority. Predicting where and how frequently faults
will rupture is at the core of earthquake risk
reduction. The similarity of these two major
fault systems had prompted joint scientic
studies for more than 30 years. Members of
the USGS team also concentrated on docu-
menting ground shaking and analyzing how
local geology controlled the patterns of dam-
age. Representatives from the National Insti-
tute of Standards and Technology documented
construction damage and drew implications for engineering design practice.
While it quickly became apparent that construction quality and the failure to
enforce local building codes were major contributors to damage, aspects of
the damage are important to the United States. These include the collapse of
reinforced concrete buildings with little or no ductility and liquefaction fail-
ures of building foundations. Representatives from the U.S. Army Corps of
Engineers visited major earth dams in the epicentral region, all of which per-
formed excellently.
The Kocaeli earthquake struck ten years after the Loma Prieta, Califor-
nia, earthquake, and ve and four years, respectively, after the devastating
urban earthquakes in Northridge, California, and Kobe, Japan. Each of these
earthquakes has affected how we mitigate the risk of earthquakes. The
Loma Prieta, California, earthquake, striking at the south edge of the San
Francisco Bay region, was the rst earthquake to cause signicant damage
in a major urban region in the United States since the enactment of the
National Earthquake Hazards Reduction Program in 1977 (U.S. Geological
Survey, 1989). It prompted intense scientic scrutiny and reinvigorated miti-
gation efforts in California; approximately $2.5 billion was spent to retrot
bridges in the State highway system. The 1994 Northridge, California,
earthquake, on the north edge of the Los Angeles metropolitan area (U.S.
Geological Survey, 1996), cost more than $12.5 billion in insured property
loss and prompted a revolution in the earthquake insurance industry in the
United States. In California, a public-private corporation, the California
Earthquake Authority, was established to provide most of the residential
earthquake insurance in California. Nationally, the insurance sector was
prompted to base insurance premiums on more rigorous estimates of nation-
wide seismic risk using models developed by earthquake scientists and
engineers. The 1995 Hyogoken-Nanbu, Japan, earthquake caused $100 bil-
lion in total property loss in the Kobe area. It further demonstrated the
potential of urban earthquakes to cause extraordinary property losses
(National Institute of Standards and Technology, 1996).


4

Now, ten years after the Loma Prieta earthquake, the
Kocaeli earthquake has again demonstrated the awesome
potential of large earthquakes and particularly the immense
human suffering that can occur when an earthquake strikes
in an urban area. In Turkey, more than 250,000 people
were forced out of their damaged homes and into huge tent
encampments. Tent cities were established both next to col-
lapsed and damaged homes and in open elds outside of
the cities (g. 1). Long lines of people waiting for food
were a constant reminder of their dependency on urban
infrastructure.
Tragedy, however, can be an opportunity if we learn
from it. The challenge now is to learn from the Kocaeli
earthquake. The contributors to this report hope that the
memory of the people who died in the Kocaeli earthquake
will inspire renewed efforts to mitigate efforts from future
earthquakes. This report summarizes implications for the
United States drawn from ndings by teams from the U.S.
Geological Survey, National Institute of Standards and
Technology, and U.S. Army Corps of Engineers Engineer
Research and Development Center.

Figure 1.

The heavy damage to residences required the use of
113,869 tents in 121 tent cities for temporary shelter.

Photographs by
Thomas L. Holzer.


5

The moment magnitude (M

W

) 7.4 Kocaeli, Turkey, earthquake struck
the Kocaeli province of northwestern Turkey on Tuesday, August 17, 1999,
at 3:02 a.m. local time. The cause of the earthquake was the sudden break-
age, or rupture, of the Earths crust along a western branch of the 1,500-km-
long North Anatolian fault system. The epicenter of the earthquake, the loca-
tion on the ground surface above where the faulting began, was at lat 40.70
N., long 29.91 E. according to the Earthquake Research Department of the
Directorate for Disaster Affairs of the Ministry of Public Works and Settle-
ment. This is about 7 km southeast of the city of Izmit and 15 km east of the
town of Glck. The earthquake nucleated at a depth of about 15.9 km
beneath the ground surface. From the point at which faulting began, the fault
rupture extended upward to the ground surface and for many tens of kilome-
ters both east and west, producing strong ground shaking along its entire
length. The total length of the fault rupture is about 110 km.
The region hit by the earthquake is the industrial heartland and the most
densely populated section of Turkey. According to ofcial government g-
ures (as of October 19, 1999) the earthquake caused 17,127 deaths and
43,953 injuries, and left more than 250,000 people homeless. An estimated
214,000 residential units and 30,500 business units either collapsed or were
lightly to heavily damaged. Figure 2 shows fatalities and damaged residen-
tial units on a map of northwest Turkey. The heavy damage to residences
required the use of 113,869 tents in 121 tent cities for temporary shelter. The
region of heaviest damage extended from Yalova and Korfez to the town of
Dzce, which is 100 km east of the epicenter. Most of the damage and fatal-
ities were in towns located on the narrow at shorelines of the Sea of Mar-
mara. There also was heavy damage and loss of life in Adapazari, which is
in the Sakarya valley. Anomalous damage, because it was isolated, occurred
in the Avcilar district of Istanbul, west of the epicenter about 80 km from the
end of the causative fault rupture.
Maximum amplitudes of accelerations recorded by the three strong
motion stations closest to the fault rupture, which were within 7.5 km of the
rupture, were smaller than is usually generated by a M

W

7.4
event. Recordings of the earthquake at two stations close to
the fault indicate it had a strong velocity pulse and was of
long duration, which were important factors in the damage
to buildings. Recorded accelerations and computed veloci-
ties and displacements from the two strong motion stations
are shown on gures 3 and 4. The velocity and the displacement records
show the impulsive character of the ground shaking.

Strong motion:

Earthquake shaking of
sufcient intensity to cause damage to
buildings and other structures.

5


6

1500
1200
900
600
300
Elevation
0
EPICENTER
Avcilar
ISTANBUL
Korfez
Izmit
Glck
Karamrsel
Yalova
Degirmendere
Adapazari
Dzce
BLACK SEA
SEA OF MARMARA
TURKEY
Black Sea
978 Deaths
23,775 Residences
8,744 Deaths
57,530 Residences
2,627 Deaths
42,181 Residences
264 Deaths
9,123 Residences
2,501 Deaths
29,829 Residences
41
40
30'
30
31
29
0 30 KILOM ETERS

Although the M

W

7.4 was the magnitude nally adopted by
all parties involved in the earthquake investigation, the wide
range in magnitudes initially reported for the earthquake played a
role in the controversy in Turkey that arose over the Turkish gov-
ernments handling of the August 17, 1999, disaster. The magni-
tudes were used by the Turkish press to symbolize perceived
government ineptitude in its postearthquake response. The USGS
National Earthquake Information Center reported a surface wave
magnitude (M

S

) of 7.8 in the hours after the earthquake, during a
time period when Kandilli Observatory, the Turkish institution
that monitors the epicentral region, was reporting that the earth-
quake had a duration magnitude (M

d

) of 6.7. In general, the M

S

of
an earthquake, which is based on the amplitudes of long-period
seismic waves, would be a more reliable measure of the destruc-
tive potential of an earthquake than would the M

d

, which is based
on the duration of shaking. However, Kandilli Observatory per-
sonnel were forced to rely on the M

d

, because many of the seis-
mometers in the Kandilli network had gone off scale, and the

Figure 2.

Major cities and towns in Turkey that experienced significant damage to residential units and loss of life during the Kocaeli earthquake.
Upper number is fatalities; lower number is lightly to heavily damaged residential units.

(Statistics from the Prime Ministry of Turkey, Crisis Manage-
ment Center, September 12, 1999).

amplitudes of seismic waves could not be accurately read.
Although a magnitude 6.7 earthquake is still recognized by seis-
mologists as having high destructive potential, there was specu-
lation in the press that reporting of the seemingly low duration
magnitude delayed recognition of the seriousness of the disaster
in the epicentral region, which had lost most of its telecommuni-
cation capability. The delay highlights the utility of digital seis-
mic monitoring networks that transmit near real time
announcements of potentially devastating earthquakes.

6


7

TIME, IN SECONDS
A
C
C
E
L
E
R
A
T
I
O
N
,

I
N

c
m
/
s
e
c
2
0 10 20 30 40 50 60
TIME, IN SECONDS
0 10 20 30 40 50 60
TIME, IN SECONDS
0 10 20 30 40 50 60
-
4
0
0
-
2
0
0
0
.
0
2
0
0
V
E
L
O
C
I
T
Y
,

I
N

c
m
/
s
e
c
-
2
0
0
2
0
4
0
6
0
8
0
D
I
S
P
L
A
C
E
M
E
N
T
,

I
N

c
m
0
5
0
1
0
0
1
5
0
2
0
0
A
B
C

Figure 3.

A

, Recorded accelerations at the Turkish National Strong Motion Network station, SKR, in Adapazari during the August 17, 1999, main
shock. Station is about 3.5 km from the surface rupture.

B

, Velocities computed from acceleration time history showing large velocity pulse.

C

, Dis-
placements computed for acceleration time history showing large displacement pulse.

Illustration by William L. Ellsworth.

7


8

TIME, IN SECONDS
A
C
C
E
L
E
R
A
T
I
O
N
,

I
N

c
m
/
s
e
c
2
0 5 10 15 20 25 30
-
3
0
0
-
2
0
0
-
1
0
0
0
1
0
0
2
0
0
3
0
0
V
E
L
O
C
I
T
Y
,

I
N

c
m
/
s
e
c
-
1
0
0
-
5
0
0
5
0
1
0
0
D
I
S
P
L
A
C
E
M
E
N
T
,

I
N

c
m
-
5
0
0
5
0
1
0
0
1
5
0
2
0
0
2
5
0
TIME, IN SECONDS
0 5 10 15 20 25 30
TIME, IN SECONDS
0 5 10 15 20 25 30
A
B
C

Figure 4

.

A

, Recorded accelerations at the Kandilli Observatory and Earthquake Research Institute strong motion station, YPT, at the Petro-Chem-
ical Plant in Korfez during the August 17, 1999, main shock. Station is about 4 km from the surface rupture.

B

, Velocities computed from acceleration
time history showing large velocity pulse.

C

, Displacements computed for acceleration time history showing large displacement pulse.

Illustration
by William L. Ellsworth.

8


9

Geologic Setting and
Earthquake History
Black Sea
Greece
AFRICAN PLATE
ARABIAN
PLATE
EURASIAN PLATE
Mediterranean Sea
Turkey
Crete

Black Sea
Greece
AFRICAN PLATE
H
e
l
l
e
n
i
c

A
r
c
ARABIAN
PLATE
EURASIAN PLATE
Aegean Sea
Cyprus
C
y
p
r
u
s
A
r
c
Mediterranean Sea
F
lo
r
e
n
c
e
R
is
e
Istanbul
Izmit
Sea of
Marmara
Caspian
Sea
Turkey
Anatolian
Block
Crete
42
39
36
33
21
27
33
39
45
51
100 0 200 KILOM ETERS

N
o
rth
Anatolian
F
ault
E
a
s
t
A
n
a
t
o
lia
n
F
a
u
lt
Mediterran
e
a
n
R
i
d
g
e

AFRICA
EUROPE
ASIA

Turkey is geologically active and frequently experiences large earthquakes.
The activity is a consequence of the country being squeezed between the north-
ward-moving Arabian and African tectonic plates and the relatively stable Eur-
asian plate (g. 5). A wedge of continental crust, which is known as the Anatolian
block and incorporates much of Turkey, is being squeezed westward away from
the collision zone between the Arabian and Eurasian plates. Movement of the
wedge is accommodated along two major strike-slip faults: the 1500-km-long
North Anatolian fault on the wedges north side and the 550-km-long East Ana-
tolian fault on the southeast side. The August 17, 1999, Kocaeli earthquake was
caused by slip along the western end of the North Anatolian fault. This major
right-lateral strike-slip fault allows the region south of the fault to move to the west
relative to the region north of the fault, at a rate of 10-20 mm/yr (Barka, 1992;
Straub and Kahle, 1995). The fault is predominantly a single strand for much of its
length with a few shorter, subparallel faults; however, it branches into three
strands in the western 650 km where the 1999 earthquake occurred (g. 6

A

). The
northernmost, and most active, of these strands traverses in part the Sea of Mar-
mara and includes the section that slipped on August 17, 1999. Surface rupture
from the 1999 Kocaeli earthquake was mainly conned to Pliocene and Quater-
nary sedimentary basins and valley ll. Some of these basins appear to be formed
between offset segments of the fault due to repeated slip of these segments.
The long written history of Turkey includes descriptions of many destructive
earthquakes during the past thousand years. Although these historical accounts do
not include descriptions of surface faulting, rupturing segments commonly can be
inferred from descriptions of damage and how far away the earthquake was felt. It
is important to know where and when the fault ruptured because scientists can use
this information to evaluate the regularity and magnitude of earthquakes that can
be expected on a specic fault segment. Furthermore, if the rate of strain accumu-
lation on the fault is known from either geologic or geodetic studies, scientists can
roughly estimate the time to the next earthquake on the same fault segment.

9

Figure 5.

The tectonic map of Turkey includes the North Anatolian fault, East Anatolian fault, and Hellenic and Florence trenches. The westward
movement of the Anatolian block results from (1) differences in rates of motion between the Arabian and African plates, (2) different directions of
motion between the Anatolian block and Eurasian plate to the north, and (3) subduction of the African plate beneath the Anatolian block at the Hel-
lenic and Florence trenches. The Arabian plate is moving to the north faster than the African plate, both relative to a stable Eurasian plate. The result
is a westward moving wedge incorporating most of Turkey.

Modified from Barka (1992) and Rockwell and others (in press).


10

Y
E
A
R

(
A
D
)
1900
1700
1500
1300
1100
900
0 250 500 750 1000 1250
DISTANCE ALONG FAULT FROM WESTERN EDGE OF GELIBOLU PENNINSULA (km)
? ?
1999
1719
1668
1509
?
1912
1967
1957
1944
1943
1951
1942 1939
1992
1949
1971
1966
1794
1784
1766
1766
1754
1666
1579
1556
1354
1344
1254
1050
1043
1035
967
Erzincan
Karliova
Istanbul
Black Sea
Black Sea
Bolu
Iznik
Erbaa
Gelibolu
NORTH ANATOLIAN FAULT ZONE
Historic Earthquakes Along the North Anatolian Fault
1999
27
31
35
39

Izmit
A
B
DZCE FAULT

Based on comparisons of reported damage and how far away
they were felt, the 1999 Kocaeli earthquake appears to be similar to
an earthquake that damaged communities in 1719. Reports of shak-
ing and descriptions of damage from the 1719 earthquake are remark-
ably similar to those observed in 1999 (Ambraseys and Finkel, 1995).
There were also two earthquakes in 1766 and one in 1754 that prob-
ably resulted from fault rupture closer to Istanbul. These earthquakes
are similar in size to those felt during the 20

th

century. In the more
remote past, however, there have been earthquakes that appear to
have ruptured more of the fault than has ruptured in any single earth-
quake during the past three centuries. In 1509, for example, an earth-
quake occurred that probably resulted from the rupture of faults from
Gelibolu on the west to Bolu on the east (g. 6

B

). Rupture of a long
section of the fault is also inferred along the eastern section during
the great earthquake of 1668. Faulting during this earthquake is
thought to have occurred over 600 km, from west of Bolu to Erzincan
(g. 6

B

). The long ruptures inferred for the 1509 and 1668 shocks
would be larger than any during the 20

th

century. Studying the
Kocaeli earthquake is important because it can provide insight into
faulting processes responsible for the variable behavior.
During the past 60 years, the North Anatolian fault has pro-
duced earthquakes along different sections of the fault in a system-
atic manner that is atypical of long faults. Beginning with the 1939
M 7.9 earthquake, which produced about 350 km of ground rupture
along the eastern North Anatolian fault, the 20

th

century has seen ten
additional moderate to large earthquakes along the fault (g. 6

B

).
Most of the earthquakes occurred sequentially in a westward pro-
gression. The 1999 Kocaeli earthquake is now the westernmost large
earthquake (on the northern strand of the fault), breaking a section
that appears, on the basis of historical reports of damage, not to have
slipped for at least 250 years. The historical record does not support
a theory that seismic behavior of the North Anatolian fault in the
20

th

century is a repetition of the pattern of earlier centuries. The
record does show that several ground-rupturing earthquakes had
occurred in close succession in the past, as with the sequences of
events in the mid-eighteenth, fourteenth, and eleventh centuries. An
incomplete historical record may hide evidence of other, longer
sequential earthquake ruptures.

10

Figure 6.

A

, Strip map showing the 1500-km-long North Anatolian fault. Relatively straight, the fault has several short, minor strands associated
with it along its eastern half and breaks into three strands to the west. The section of the fault that ruptured to the surface during the 1999 Kocaeli
earthquake is shown in red. In 1999, the fault ruptured along the northernmost, generally east-west main strand, as well as along a northerly strand
called the Dzce fault. Modified from Barka (1992) and Rockwell and others (in press).

B

, Historical earthquakes located along the fault where they
are believed to have ruptured the ground surface. The 1999 rupture appears to be similar to the 1719 rupture. Solid lines indicate an earthquake
known to have ruptured the ground, long-dashed lines indicate highly probable surface-rupturing events, and short-dashed lines are shown when
the earthquake is believed to have ruptured the ground but that interpretation of the historic record has less certainty. Before 1719, the same section
of the fault slipped as part of a longer rupture. There is no record in the 20

th

century of the kind of event that produced the long rupture in 1668. Gaps
in the historic record of earthquakes may indicate no seismic activity or simply that there is no record of the seismicity. Because few population
centers are in the central portion of the fault, the lack of seismic record may be accentuated.

Compiled from Ambraseys (1970), Ikeda and others
(1991), Ambraseys and Finkel (1995), Barka (1996), Stein and others (1997), and Rockwell and others (in press).


11

Earthquake Forecasts
1999
1967
1944 1943
1939
1942
1951
1957

The sequential westward progression of M

W

>6.7 earthquakes along the
North Anatolian fault during the 20

th

century has prompted study of how one
large earthquake in Turkey may set up the next one (g. 7) (Stein and others,
1997; Nalbant and others, 1998). One recent investigation consisted of calcu-
lations of the earthquake-induced changes in stress on adjacent fault seg-
ments (Stein and others, 1997). Stein and others (1997) showed that 9 out of
the 10 observed ruptures were brought closer to failure by the preceding
shocks; they estimated that the shocks are equivalent to about 20 years of
stress that would typically build up along the North Anatolian fault. The
stress change also affects the change in the probability of future events. The
permanent change in probability occurs because increasing (or decreasing)
the stress on a fault shortens (or lengthens) the time to bring a segment to fail-
ure. There is also a transient change in probability that strongly amplies the
effect of the stress change for periods of about a decade. For the North Ana-
tolian fault, the computed stress change yielded an average threefold gain in
the ensuing earthquake probability. Before the Kocaeli earthquake, Stein and
others (1997) calculated that stresses at several isolated sites along the fault
had been increased by prior events. They estimated a 15 percent probability
of a M

W

>6.7 earthquake east of Erzincan and a 12 percent probability for a
large event south of the city of Izmit during the interval from 1997 to 2027 on
the Izmit segment of the North Anatolian fault. The August 17, 1999, Kocaeli
earthquake ruptured the Izmit segment.
Application of this modeling technique to evaluate the effect of the
Kocaeli earthquake on other faults indicates that stress was transferred to
nearby fault segments both to the east and west (g. 8). To the east, the area
of increased stress included the Dzce fault on which the November 12,
1999, M

W

7.1 earthquake occurred. To the west, both the 80-km-long Yalova
segment, southeast of Istanbul, and the North Boundary fault, immediately
south of Istanbul, are closer to failure. A cluster of aftershocks of the 1999
Kocaeli earthquake are located 50 km west of the end of the Kocaeli rupture
on the Yalova fault segment, suggesting that the calculated stress changes
have increased the rate of seismicity on the Yalova fault. There also have been
scattered aftershocks on the North Boundary fault.
The November 12, 1999, earthquake also occurred at a location along
the North Anatolian fault system where strain appears to have been aseismi-
cally accumulating. Fault slip during earthquakes located between and
including the 1939 Erzincan and 1999 Kocaeli events was at a minimum
along the interval that includes the Dzce fault (g. 7

B

,

C

), where the
November 12 event produced as much as about 5 m of slip. The ongoing rel-
ative movement between the Eurasian plate and Anatolian block of about 20
mm/yr causes strain to accumulate steadily along the North Anatolian fault.

Figure 7 (facing page, top).

The North Anatolian fault, Turkey.

A

, Loca-
tion of main fault and related strands;

B-D

are projected relative to its
pole of rotation (where the dashed lines converge).

B

,

C

, The westward
progression of earthquake ruptures is associated with fault slip of 2-6 m
in each event, with 4 m as average. The August 1999 event is seen to be
typical in extent along strike and in slip, although the 1967 and 1999
events significantly overlap on roughly parallel fault strands of the North
Anatolian system.

D,

Stress build-up on the North Anatolian faults asso-
ciated with the 1939-1967 westward progression of earthquakes and the
contribution of steady stress accumulation. Although the stress dropped
along 750 km of the North Anatolian fault, it was calculated to have been
increased at the site of the Kocaeli earthquake (Stein and others, 1997).
The August 17, 1999, ground rupture is shown in red.

Modified from Stein
and others (1997).

11

Figure 8 (facing page, bottom).

Map showing stress transferred to all
known or suspected faults in the Sea of Marmara by the 1999 Kocaeli
earthquake. Note that off-fault aftershocks tend to locate where the
Coulomb stress is calculated to have risen as a result of the 1999 Ko-
caeli earthquake rupture. The November 12, 1999, M

W

7.1 occurred in
the area of increased stress east of the August surface rupture. The
Yalova segment and the North Boundary fault, B, a probable site of a
1963 M

W

6.3 earthquake are identified. Sites of other large shocks in
the Sea of Marmara are also identified.

Prepared by Ross Stein.


12

Izmit
Istanbul
1999
0 -2 -1 1 2
Coulomb Stress Change, in Bars
Erzincan
DISTANCE, IN KILOMETERS
D
I
S
T
A
N
C
E
,

I
N

K
I
L
O
M
E
T
E
R
S
0 200 400 600 -400 -200
-100
100
0
1967
1992
0 200 400 600 -400 -200
Izmit
Istanbul
Karamrsel
1999
Erzincan
1967
1992
1971
1957
1944
1951
1943
1966
1949
1992
1942
1939
BLACK SEA
BLACK SEA
Hendek
Yalova
Iznik Geyve
Sapanca
Dzce
Alaca
Ovacik
Yedisu
EAST
ANATOLIAN
NORTHEAST
ANATOLIAN
NORTH
A
N
AT
OLIAN
NORTH ANATOLIAN
D
I
S
T
A
N
C
E
,

I
N

K
I
L
O
M
E
T
E
R
S
-100
100
0
0 200 400 600 -400 -200
August
1999
1967
1944
1943
1939
1942
1951
1957
1992
1949
1971
1966
C
U
M
U
L
A
T
I
V
E

R
I
G
H
T
-
L
A
T
E
R
A
L

S
L
I
P
,
I
N

M
E
T
E
R
S
8
6
4
2
0
TURKEY
A
B
C
D
1999
Ground Rupture
1912
M>5
M>4
M>3
M>2
Kocaeli Aftershocks
-1 0 1
Coulomb Stress Change, in Bars
Reduced Hazard Increased Hazard
Istanbul
BLACK SEA
SEA OF
MARMARA
1935
Y
a
lo
v
a
1963
1967
1943?
1953
1964
B
North Boundary Fault
EPICENTER

12

0.9
5.1
3.8
2.1
2.0
2.7
0.2
0.7
0.1 0.2 1.2
1.3
1.5
0.7
0.5 0.2
3.0
2.7 3.2 2.2
4.0
2.0
1967 Mudurnu Valley surface rupture
W
estern D
uzce fault
Sakarya section
EPICENTER
40
45"
41
29
15' 29
30' 29
45' 30
30
30' 30
45' 31
31
15'
0 8 16 KILOM ETERS
Surface Faulting
The August 17, 1999, Kocaeli earthquake produced right-lateral
displacements or slip on surface ruptures over a distance of at least 110 km
onshore in western Turkey (fig. 9). Rupture occurred along the North
Anatolian fault to the west of the epicenter near Izmit, as well as to the east
toward Akyazi, and along a northeast-trending fault toward Dzce. Thus,
the fault ruptured in both east and west directions from the epicenter. The
August 1999 surface rupture is characterized by an east-west trending zone
of right-stepping fault strands, which express a component of extension at
the western end of the northern strand of the North Anatolian fault system
(Armijo and others, 1999). Maximum slip of approximately 5 m was
mapped 31 km to the east of the epicenter; slip was typically 2.5 to 4.5 m
right lateral at the surface. Slip decreased gradually from the maximum to
the east, but more abruptly to the west and at locations where the surface
trace was stepping over to a new trace (for example, at Lake Sapanca (fig.
9)). The slip also diminished abruptly to the east where it paralleled the
ground rupture caused by the 1967 Mudurnu Valley earthquake. The fault
zone was typically a narrow zone of cracks (5 to 25 m wide) that either
followed or formed parallel sets of faults oblique to the trend of faulting.
The surface expression of faulting included tension cracks and fissures with
localized positive relief. The August 1999 surface rupture appeared to be
located in the general vicinity of previous surface faulting events, as a pre-
existing scarp was present at several locations where displacement across
the rupture had a vertical component.
13 14
Figure 9. Preliminary map showing surface rupture along the North Anatolian fault zone during the Kocaeli earthquake. The rupture is marked
in red with selected values of offset (in meters) along the fault. The western end of the ground rupture associated with 1967 Mudurnu Valley earth-
quake is also shown where it parallels the 1999 surface rupture. Surface rupture and offsets was mapped in detail by four groups of geologists
from the USGS, the Southern California Earthquake Center (SCEC), the Institut de Physique du Globe (IPGP), France, and Istanbul Technical
University (IT). Each group focused on one of the four sections of the fault that ruptured (Barka and others, 1999).
Figure 10. Tectonic scarp along the normal fault in Glck in a cobblestone pavement with both vertical (~1 m)
and horizontal (~0.3 m) components of offset. Photograph by Robert Langridge.

15

Areas under water in blue
Coastal Changes Due to Kocaeli Earthquake
0 1 KILOM ETER

The westernmost extent of faulting found on land was in
Glck where a seawall was offset approximately 4 m. To the
west, the rupture crossed a small bay at Degirmendere and the
nearby shore had subsided. Surface rupture was not reported on
the Hersek Peninsula, and it is likely that offshore rupture of the
sea oor stopped west of Degirmendere. A south-facing scarp
near the tip of the Hersek Peninsula may be part of the next east-
west trending segment of the North Anatolian fault to the west.

In Glck, east-west trending strike-slip faulting, with off-
sets of as much as 4.5 m, extended through the waterfront and
port region. Part of the strike-slip faulting transfers eastward into
a N. 70

W. trending zone of predominantly normal faulting. Ver-
tical displacements of as much as 2.4 m were recorded on the
normal fault in Glck. The fault also had a modest right-lateral
strike-slip component, 1.2 m at maximum, 0.5 m on average (g.
10). The structural relationship, the 4-km-long scarp with pre-
existing geomorphology along the fault trace, and subdued
topography around the bay suggest that this scarp is related to
tectonic faulting rather than landsliding or delta-front collapse.
Trenching of the normal fault conducted immediately after

the
Kocaeli earthquake revealed four previous seismic events, two
dated at A.D. 1509 and A.D. 1719 and two undated older events.
Another shorter normal fault lies to the east of and parallel to the
normal fault in Glck. Additional strike-slip faulting was not
observed on the south side of Izmit Bay (B. Meyer, oral com-
mun., 1999). The normal faulting in the Glck area caused
regional subsidence that submerged part of the coast (g. 11).
Approximately 4 km of coastline on the south shore of Izmit Bay
subsided about 2 to 3 m. Many structures built near the shore
were inundated by the sea (g. 12).
The rupture continued onshore east of Izmit Bay where it
was characterized by an impressive mole track and caused dam-
age to buildings and other cultural features (gs. 13, 14, 15). The
fault right laterally offset the Trans European Motorway (TEM)
near Tepetarla by about 3 m and extended to the northwest corner
of Lake Sapanca. The lake lls a pull-apart basin created by long-
term motion on the North Anatolian fault. The fault probably fol-
lows the north side of Lake Sapanca to near Esme, which suffered
extensive damage from liquefaction. Projection of the fault trace
through Lake Sapanca requires a right stepover of as much as 2
km to where the fault is next seen to the east onshore.
The 26-km-long Sakarya section of the ground rupture from
Lake Sapanca to Akyazi was the focus of mapping efforts by the
team of USGS geologists and is shown in detail on gure 16
(Fumal and others, 1999). Southeast-trending tensional cracks
near the lakeshore connected to an east-west trending mole track

Figure 11.

Satellite image of Glck area showing zones of coastal submergence (blue). Strike-slip faulting occurred through the port area (upper
left) and oblique faulting occurred on the Glck (normal) fault.

Copyrighted image by CNRS-IPGP and SPOT image; used with permission.

15


16

Figure 12.

Submergence of shoreline in Glck caused by tectonic sub-
sidence associated with normal faulting.

Upper photograph by Robert
Langridge; lower photograph by Thomas L. Holzer.

Figure 13.

Mole track along ground rupture near Izmit.
Right-lateral strike-slip displacement was about 3 m. Positive
topographic relief of the mole track was localized and did not
represent a component of the motion.

Photograph by Heidi
Stenner.

17 18
Figure 16. The Sakarya section of the 1999 surface rupture mapped by USGS geologists. Site localities where measurements were taken shown as
dots; data are unpublished but available by request from R.M Langridge or Heidi Stenner, USGS.
Figure 14. Offset canal near
Izmit. Strike-slip displacement
across the whole zone here
was more than 2.5 m and
included a vertical component
(middle left). Photograph by
Heidi Stenner.
Figure 15. Surface
rupture extending
through boundary wall
and collapsed apart-
ments, to the east of
Izmit. Strike-slip dis-
placement here was
about 3 m. Photograph by
Heidi Stenner.

19

with less than 0.5 m of right-lateral displacement. The rupture
continued eastward crossing the TEM and railway tracks. Right-
lateral offset increased from less than 0.5 to 4.3 m over a distance
of 1.5 km. A secondary trace was found that followed the south-
ern range front of the lake from the southeast edge and exhibited
as much as 0.5 m of right-lateral displacement before it merged
into the main zone of faulting. The maximum right-lateral slip
measurement of 5.10.2 m was recorded near Ariye in a grove
of offset poplar trees (g. 17). Offsets on other cultural features
nearby conrm that there was a large amount of slip in this region
(gs. 18, 19). Slip decreased uniformly to the east; right-lateral
slip at the Sakarya River was 3.30.2 m.

Stepover zones between east-west trending fault strands
were characterized by a change in strike of the ground rupture,
widening of the fault zone with an increase in the number of
strands, a vertical component of offset, and a decrease in the total
slip recorded. Along the Sakarya section, stepovers between fault
strands, greater than 1 km in length, were typically 100 m wide
(g. 16). East of the Sakarya River, right-lateral slip decreased to
approximately 2 m along the section of the rupture that parallels

Figure 17.

Offset poplar trees (2.7 m right lateral) along
the surface rupture near Resitbey. The rupture traversed
many poplar groves, cornfields, and arable land along the
Sakarya River plain.

Photograph by Heidi Stenner.

Figure 18.

Offset wall near Arifiye. The right-lateral displacement is
4.58 m.

Photograph by Heidi Stenner.


20

the surface rupture associated with the 1967 Mudurnu Valley
earthquake (g. 9). At Kanliay, 76 cm of total slip was recorded
near the west end of the 1967 rupture (Ambraseys and Zatopec,
1969); at the same longitude, 7.5 km to the north, 2.3 to 2.9 m was
measured along the 1999 rupture (g. 20). Evidence was found
for slip on the 1967 fault trace that was triggered by the 1999
earthquake. East of Kanliay on the 1999 rupture, slip decreased
to less than 0.5 m and the rupture stepped to the left. Surface rup-
ture was recognized again at Kazanci. From Kazanci, the rupture
trended N. 75

W. toward Akyazi, and was characterized by right-
stepping N. 37

W. to N. 55

W. trending tension gashes; slip
increased from 0.04 m to 0.7 m on this strand, and died out to the
east of Akyazi.
A 6-km gap in the surface rupture occurred from Akyazi to
Trkbeynevit where surface rupture was observed again along the
N. 65

E. trending western Dzce fault (g. 9). The Dzce fault
follows the northwest-edge of an uplifted bedrock block into a
fault-strike valley toward Glyaka. Despite the change in strike,
this segment had predominantly right-lateral displacements along
its 30 km length. Slip increased from less than 0.5 m to a maximum
of 1.5 m near the east end of the valley. Slip then decreased toward
the town of Glyaka. The surface rupture terminated along the
southern margin of a small pull-apart called Eften Gl or Lake
Eften. East of Glyaka, the western Dzce fault steps to the right
twice to form two short fault segments that bound the range front
of Eften Gl, and marks the termination of faulting (Hartleb and
others, 1999). The nal strand had right-lateral slip of as much as
0.2 m. From north to south, the western Dzce fault also parallels
the ground rupture during the 1967 Mudurnu Valley earthquake.

Figure 19.

Road offset right laterally 4.66 m near Arifiye.

Photograph by Robert Langridge.

Figure 20.

Complex surface rupture pattern with linking en echelon
cracks seen near the Amalik canal site. Right-lateral strike-slip
displacement here amounted to 2.26 m and was 2.40 m in the dike of the
canal, seen in the distance.

Photograph by Robert Langridge.


21

Ground Shaking
Implications of
Aftershock Observations
Main-Shock
Observations
0 60 KILOM ETERS
Strong motion
station
Active fault
N

On-scale recordings of strong ground shaking close to the fault rupture
are important for both understanding causes of earthquake damage and the
physics of fault rupture. Approximately 38 recordings of strong motion were
made of the August 17, 1999, Kocaeli earthquake main shock. The recordings
were made by four of the ve institutions in Turkey that operate either strong
motion networks or small arrays. Much of the strong motion data that was
collected are available on the Internet. The ve institutions, which are
arranged in order of the size of their networks, and their Internet sites are as
follows:
1. National Strong Motion Network (NSMP), operated by the Earth-
quake Research Department, Directorate for Disaster Affairs of the
Ministry of Public Works and Settlement (ERD)
<http://angora.deprem.gov.tr/>
2. Kandilli Observatory and Earthquake Research Institute (KOERI)
<http://www.koeri.boun.edu.tr/earthqk/earthqk.html>
3. Istanbul Technical University (IT) <http://www. itu.edu.tr/>
4. Public Water Works (DSI) <http://www.dsi.gov.tr/)>
5. Middle East Technical University (METU)
<http://www.metu.edu.tr/home/wwweerc/> and
<http://www.metu.edu.tr/home/wwwdmc/>

Figure 21.

Modified map showing the National Strong Motion Network of Turkey and major active faults, including the North Anatolian fault.

Modified from Earthquake Research Department, Directorate for Disaster Affairs, Ministry of Public Works and Settlement.

21


22

41
41
30'
30
31
29
0 50 KILOM ETERS
BLACK SEA
SEA OF MARMARA
ISTANBUL
Izmit
Adapazari
0.19
0.25
0.09
0.17
0.12
0.19
0.07
0.06
0.11
0.21
0.26
0.32
0.12
0.23
0.41
0.04
0.05
N
EPICENTER

Station L,
in g
T,
in g
L
(+)
T
(+)
V,
in g
Lat N. Long E. Network
operator

IZT 0.171 0.225 S E 0.146 40.790 29.960 ERD
SKR * 0.407 S E 0.259 40.737 30.384 ERD
DZC 0.374 0.315 W S 0.480** 40.850 31.170 ERD
IST 0.061 0.043 S E 0.036 41.080 29.090 ERD
GBZ 0.264 0.142 N W 0.199 40.820 29.440 ERD
CEK 0.118 0.190 N W 0.050 40.970 28.700 ERD
IZN 0.092 0.123 S E 0.082 40.440 29.750 ERD
BRS 0.054 0.046 S E 0.025 40.183 29.131 ERD
YPT 0.230 0.322 W N 0.241 40.763 29.761 KOERI
ATS 0.252 0.180 N W 0.081 40.980 28.692 KOERI
DHM 0.090 0.084 S W 0.055 40.982 28.820 KOERI
YKP 0.041 0.036 S W 0.027 41.081 29.007 KOERI
FAT 0.189 0.162 S E 0.131 41.054 28.950 KOERI
ARC 0.211 0.134 N W 0.083 40.83 29.36 KOERI
HAS 0.056 0.110 S E 0.048 40.869 29.090 KOERI
MCK 0.054 0.070 N W 0.038 41.065 28.990 IT
ZYT 0.120 0.109 N W 0.051 40.986 28.908 IT
MSK 0.054 0.038 N W 0.031 41.104 29.010 IT
ATK 0.103 0.168 N W 0.068 40.989 28.849 IT

Table 1.

Coordinates and peak accelerations
recorded during the 1999 Kocaeli, Turkey, earth-
quake at strong motion stations that recorded sig-
nificant shaking.

[

L, longitudinal, T, transverse, V, vertical; the compo-
nents L and T are instrument components and do not
correspond to north-south and east-west automatically.
Refer to each network for the correct orientation of each
horizontal component. * L, component did not function;
**, based on a single spike (actual value may be
smaller)]

Figure 22.

Map showing locations of peak horizontal accelerations recorded at strong motion stations in and near the epicentral area. The network
stations included are from ERD, IT, and KOERI.

Base map courtesy of BKS Surveys Ltd., N. Ireland.


23

0 20 40 60 80 100 120 140
0
-0.4
-0.2
0
0.2
0.4
TIME, IN SECONDS
TIME, IN SECONDS
0
0.2
-0.2
A
C
C
E
L
E
R
A
T
I
O
N
,

I
N

g
Longitudinal (malfunction)
Transverse
Vertical
0 5 10 15 20 25 30 35 40
-0.2
0
0.2
A
C
C
E
L
E
R
A
T
I
O
N
,

I
N

g
Transverse
Vertical
A
B
-0.4
-0.2
0
0.2
-0.2
-0.4
-0.6
-0.8

The 38 main-shock recordings included 24 from ERD, 10
from KOERI, 4 from IT, and 3 sets of structural response
records, 2 from KOERI and 1 from METU. DSI did not retrieve
any records from its dams in the earthquake region. The struc-
tural records retrieved by KOERI include the structural response
arrays at Sleymaniye Mosque and Aya Sofya (Hagia Sophia)
Museum <http://www.koeri.boun.edu.tr/earthqk/ earthqk.html>.
Partial records were obtained by METU from an instrumented
six-story building in Gerede (P. Glkan, written commun., 1999).
The largest peak acceleration at the basement of the six-story
building was 0.035 g.
Although the number of instruments deployed in the epicen-
tral area may seem sparse, the number is in part limited by a
broader effort by ERD to include every major town in Turkey in
the network. Figure 21 shows the NSMP network operated by
ERD, which is the largest network operation in Turkey.
Table 1 compiles peak accelerations for each of the three
components of accelerometers at 19 strong motion stations
located in or near the epicentral region. Figure 22 is a map of the
peak horizontal acceleration recorded at these stations. Most of
the stations are housed in buildings, some of which are more than
two stories high. Standard practice in the United States when com-
piling free eld recordings is to ignore recordings from buildings
that are more than two stories high because it is believed that the
structural response of these buildings may modify the recording
from free eld conditions. Soil conditions beneath the stations in
Turkey also have not been systematically documented.

23

Figure 23.

A

, Time histories of acceleration for 140 s at SKR; note that longitudinal component malfunctioned. The plot shows several events:

B

,
time history with first 40 s of the acceleration record;

C

, the cumulative sum of the squared acceleration values; and

D

, effective duration of signif-
icant strong motion during the main shock, based on time between 5 and 95 percentiles of the cumulative sum, is 4.5 s.

Illustration by Mehmet elebi.


24

TIME, IN SECONDS
0 5 10 15 20 25 30 35 40
x 10
5
0
0
-100
-200
-300
100
200
300
1
0.8
0.6
0.4
0.2
0
2
4
6
S
U
M

O
F

A
C
C
E
L
E
R
A
T
I
O
N

S
Q
U
A
R
E
D
,

I
N

(
c
m
/
s
2
)
2
A
C
C
E
L
E
R
A
T
I
O
N
,

I
N

c
m
/
s
2
N
O
R
M
A
L
I
Z
E
D

S
U
M

O
F

A
C
C
E
L
E
R
A
T
I
O
N
Transverse
Vertical
C
TIME, IN SECONDS
0 5 10 15 20 25 30
D
95 %
5 %

The time history of acceleration recorded at SKR (Adapaz-
ari in Sakarya Province), a ERD station, is shown in gure 23

A

.
The station is on soft rock and about 3.5 km from the surface
trace of the fault. The record reveals at least two aftershocks
immediately followed the main shock. Figures 23

B

,

C

show
only 40 s of the record and gure 23

D

shows the cumulative
sum of the square of the acceleration superimposed on the
acceleration time history. This sum has been used by Trifunac
and Brady (1975) to dene an effective earthquake duration.
The effective duration is the time between the 5 and 95 percen-
tile values of the cumulative sum. This yields an effective dura-
tion of the main shock at SKR of approximately 4.5 s. This
duration is different than the duration felt and reported by
humans. People typically can sense ground shaking at levels of
less than 0.01 g. The main shock contributes approximately 97
percent of the signicant ground shaking of the two shocks
within the 40 s of the record as inferred from the cumulative
sum of the square of the acceleration.
The time history of acceleration recorded at YPT (Petro-
Chemical Plant in Korfez), a KOERI station, is shown in gure
24

A

. The station is located on alluvium. The time history reveals
two distinct earthquake pulses. Figure 24

B

shows the cumulative
sum of the square of acceleration, indicating an effective duration
of approximately 15 s for the main shock. The main shock con-
tributes approximately 85 percent of the signicant ground shak-
ing caused by the two events. The YPT instrument is housed on
the rst oor of a three-story building (g. 23

C

).

24


25

0 20 40 60 80 100 120
TIME, IN SECONDS
0 20 40 60 80 100 120
TIME, IN SECONDS
A
C
C
E
L
E
R
A
T
I
O
N
,

I
N

g
Longitudinal
Transverse
Vertical
0
0.5
1
1.5
2
x 10
6
S
U
M

O
F

A
C
C
E
L
E
R
A
T
I
O
N

S
Q
U
A
R
E
D
,

I
N

(
c
m
/
s
2
)
2
Longitudinal and transverse
Vertical
A
B
0.2
0.2
0
-0.2
-0.4
0
-0.2
0.2
0
-0.2
C

25

Figure 24.

A

, Time histories of acceleration
at YPT showing a second event approximately
33 s after the main shock.

B

, The main shock
contributes approximately 85 percent of the
total significant strong motion and its effective
duration was about 15 s. Illustration by Meh-
met elebi.

C

, Three-story building that hous-
es the YPT accelerograph. Instrument is inside
on first floor next to entrance.

Upper photo-
graph by Mehmet elebi, lower photograph by
Thomas L. Holzer.


26

0 20 40 60 80 100 120
TIME, IN SECONDS
0 20 40 60 80 100 120
TIME, IN SECONDS
Longitudinal
Transverse
Vertical
0
0.02
0.04
0.06
0.08
S
U
M

O
F

A
C
C
E
L
E
R
A
T
I
O
N

S
Q
U
A
R
E
D
,

I
N

(
c
m
/
s
2
)
2
Longitudinal
Transverse
Vertical
A
B
0
-0.05
0
0.05
0
0.05
A
C
C
E
L
E
R
A
T
I
O
N
,

I
N

c
m
/
s
e
c
2
-0.05
-0.1

26

Figures 25

A

,

B

show the acceleration time history and the
cumulative sum of the squared accelerations of the Mecid-
iyeky (Istanbul) record (MCK), an IT station. This station is
located on rock.
Figure 26 shows actual and normalized (with respect to
zero-period accelerations) pseudo-acceleration response spec-
tra, for north-south and east-west directions, respectively, for 5
stations, DZC, IZT, MCK, SKR, and YPT. Four of the stations
(DZC, IZT, SKR, YPT) are in the epicentral area, two of which,
(DZC, SKR) are near areas with heavy damage in the eastern
part of the epicentral area. The fth station (MCK) is in Istan-
bul. Three stations (DZC, IZT, YPT) are on alluvium whereas
two stations (SKR, MCK) are on either stiff soil or soft rock.
The response spectra reveal different resonant periods at each
station. The normalized response spectra indicate that both YPT
and DZC have long resonant periods or low-frequency
responses. For comparison, the design spectra for the current
Turkish Building Code for soil and alluvial site conditions are
also shown in gures 26

C

,

D

(Aydinoglu, 1998). For periods
between 0.1 and 1 s, the design spectra, which are similar to
those used in the United States, were either nearly or actually
exceeded during the Kocaeli earthquake. The majority of the
structures in the epicentral area were built to a lower design
spectra before the 1998 code was implemented. Thus, the
strength of many structures would have been insufcient to
resist the forces generated by the Kocaeli earthquake even if
the structures had been constructed according to existing codes.

Figure 25.

A

, Time histories of acceleration at MCK in Istanbul. The plot shows several events.

B

, the cumulative sum of the squared acceleration
values indicates the main shock contributed at least 90 percent of the total significant strong motion and the effective duration of strong motion was
36 s.



27

0
0.5
1
S
P
E
C
T
R
A
L

A
C
C
E
L
E
R
A
T
I
O
N
,

I
N

g
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
0
0.5
1
1.5
PERIOD, IN SECONDS
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
PERIOD, IN SECONDS
S
P
E
C
T
R
A
L

A
C
C
E
L
E
R
A
T
I
O
N
,

I
N

g
NORTH-SOUTH
EAST-WEST
A
B
YPT
IZT
SKR
MCK
DZC
YPT
IZT
SKR
MCK
DZC

27

Figure 26.

Actual pseudo-acceleration response spectra (

A, B

) and normalized (with respect to zero-period acceleration) response spectra (

C,
D

) for 5 stations. The normalized response spectra also show the response spectra of the 1998 Turkish Building Code for stiff and alluvial soil site
conditions.



28

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
0
1
2
3
4
0
1
2
3
4
5
NORTH-SOUTH
EAST-WEST
PERIOD, IN SECONDS
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
PERIOD, IN SECONDS
N
O
R
M
A
L
I
Z
E
D

S
P
E
C
T
R
A
L

A
C
C
E
L
E
R
A
T
I
O
N
N
O
R
M
A
L
I
Z
E
D

S
P
E
C
T
R
A
L

A
C
C
E
L
E
R
A
T
I
O
N
C
D
YPT
IZT
SKR
MCK
DZC
Stiff Soil
Alluvial Soil
YPT
IZT
SKR
MCK
DZC
Stiff Soil
Alluvial Soil

28


29

0 10 KILOM ETERS
Isoseismal
line
Active fault
Railroad
N

The small number of strong-motion stations in the epicentral
region that recorded the main shock precludes a detailed mapping
of ground motions. However, the areal extent of strong shaking
can be inferred from mapping damage and reports of felt shaking.
The ERD of the Ministry of Public Works and Settlement issued
a preliminary Modied Mercalli intensity map in late September
1999, a month after the main event (g. 27). The Modied Mer-
calli scale uses observations of damage to infer the intensity of
shaking (Wood and Neumann, 1931). About 150,000 km

2

was
shaken at Modied Mercalli intensity VII or greater, intensities
that prompt people to run outdoors.

Figure 27.

Preliminary isoseismal map of Modified Mercalli intensities (Roman numerals) of the August 17, 1999, Kocaeli earthquake.

Modified from
Earthquake Research Department, Directorate for Disaster affairs, Ministry of Public Works and Settlement.

29


30

100 50 0
100 50 0
A
C
C
E
L
E
R
A
T
I
O
N
,

I
N

g
0.2
0.3
0.4
0.5
0.0
0.1
A
C
C
E
L
E
R
A
T
I
O
N
,

I
N

g
0.2
0.3
0.4
0.5
0.0
0.1
IZT
SKR
DZC
IST
GBZ
CEK
IZN
BRS
ATS
YKP
FAT
ARC
HAS
MCK
ZYT
MSK
ATK
YPT
IZT
SKR
DZC
GBZ
CEK
IZN
YPT
ATS
ARC
HAS
IST
BRS
YKP
FAT
MCK
ZYT
MSK
ATK
DHM
DHM
A
B

Peak horizontal accelerations from the Kocaeli earthquake
for the 19 stations compiled in table 1 are compared to values
predicted by the empirical attenuation relation proposed by
Boore and others (1994) in gure 28. Their attenuation relation,
which predicts how peak accelerations decrease with distance
from the surface rupture, is based on regression analyses of peak
accelerations recorded in western North America. The compari-
son suggests that observed peak horizontal accelerations were
below average within at least 7.5 km of the fault rupture for an
M

W

7.4 earthquake. At greater distances from the fault, the
values were more typical of those to be expected for an M

W

7.4
earthquake although values from recorded at distances from 10 to
50 km from surface rupture also are slightly on the low side. One
of the factors in the regression is the soil condition at each record-
ing site. Peak ground motions are generally higher at soil sites
than rock sites. Because site conditions are poorly documented at
most of the strong motion stations in Turkey, values are com-
pared to attenuation curves for two different site conditions. Fig-
ure 28

A

compares the observed values to an attenuation curve for
sites underlain by materials with an average shear wave velocity
of 760 m/s in the upper 30 m. Figure 28

B

compares observed val-
ues to an attenuation curve for site conditions with an average
shear wave velocity of 360 m/s in the upper 30 m beneath the sta-
tion. The 760 m/s velocity is the boundary between rock and very
dense soil/soft rock; the 360 m/s velocity is the boundary
between very dense soil/soft rock and stiff soil (Federal Emer-
gency Management Agency, 1998). Based on eld inspection of
the three stations that are closest to the surface rupture, one
(SKR) is estimated to be near the 760 m/s boundary and two (IZT
and YPT) are estimated to be near the 360 m/s boundary.

Figure 28.

Values of peak horizontal acceleration recorded during the
Kocaeli earthquake as a function of distance from the fault compared to
values predicted by Boore and others (1994) for a M

W

7.4 earthquake in
Western North America.

A

, Values predicted for sites with average
shear wave velocities of 760 m/s.

B

, Values predicted for sites with aver-
age shear wave velocities of 360 m/s. Uncertainty is shown by the one
standard deviation curves (dashed lines).

Illustration by Selcuk Toprak.


31

Ground Shaking
Implications of
Aftershock Observations
Main-Shock
Observations

As was also the case in Turkey, the number of on-scale recordings of the
main shock for most large earthquakes is insufcient to evaluate fully the role
of geology in localizing damage. Large earthquakes, however, are usually fol-
lowed by abundant aftershocks and the Kocaeli earthquake was typical in this
regard. Because seismic waves from aftershocks and main shocks travel sim-
ilar paths from source to site, recordings of aftershocks can be used to under-
stand better how geologic conditions impacted local variations in ground
shaking and damage during the main shock at sites where the main shock was
not recorded. Thus, one of the goals of the team of USGS scientists that went
to Turkey immediately following the August 17 earthquake was to use 16 por-
table, digital seismographs and supporting equipment to help assess how the
local geology related to selected damaged areas (g. 29). Data from the initial
visit, from August 24 to September 2, are discussed here. The main objectives
of USGS scientists were to study both (1) site effects by comparing aftershock
ground motions recorded at sites inside and outside the damage areas, and (2)
propagation of the seismic waveeld recorded on small arrays.

31

Figure 29.

USGS scientists deploy an antenna for a portable seismic recording sys-
tem at Dilovasi. Antenna receives time signals from a satellite and allows the timing
on the portable seismographs to be synchronized with each other.

Photograph by
Thomas L. Holzer.


32

Avcilar West Istanbul Dilovasi Korfez Adapazari
20
40
60
80
100 km
4.1
4.6
5.2
4.8
N
Z N E Z N E Z N E Z N E Z N E Z N E Z N E Z N E Z N E Z N E Z N E Z N E Z N E Z N E Z N E
T
I
M
E
,

I
N

S
E
C
O
N
D
S
40
38
36
34
32
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
28
30'
41
00'
41
15'
40
45'
40
30'
40
15'
28
45' 29
00' 29
15' 29
30' 29
45' 30
00' 30
15' 30
30' 30
45' 31
00' 31
15' 31
30'
A
B

32

Figure 30.

A

, Map of Kocaeli earthquake aftershock zone and sites of USGS portable, digital seismographs (blue diamonds). Black concentric cir-
cles centered on the M 5.2 aftershock epicenter are in 20-km increments. Red circles are aftershock locations and their radii are proportional to their
size; the magnitudes of the four largest events are annotated.

B

, The corresponding three components of ground motion for the M 5.2 aftershock
recorded at the USGS sites are shown as a pseudo-record section of 40-s duration; Z, vertical, N, north-south, E, east-west. The amplitudes of each
trace are normalized to the peak value of that trace.

Illustration by Edward Cranswick.

33 34
Figure 31. Rescue effort 4 days after the main shock at a collapsed building in area of concentrated damage in Avcilar. Photograph by
Thomas L. Holzer.

35

Dilovasi Adapazari
Fill Rock Toyota West Toyota North
Z R T
Korfez Small-Aperture Seismic Array
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
0
4
8
12
16
20
24
28
32
36
40
44
48
52
56
60
A
B
T
I
M
E
,

I
N

S
E
C
O
N
D
S
T
I
M
E
,

I
N

S
E
C
O
N
D
S

Figure 30

A

shows the locations of USGS instruments and
epicenters of aftershocks recorded by at least one station during
the initial visit. The Avcilar district, an area of severe damage, is
located west of the Istanbul airport and about 80 km from the end
of the main-shock seismic source zone. The damage in Avcilar
(g. 31) was surprising, given its distance from the main shock
and the fact that it is surrounded by areas that sustained little
damage. Three instruments were located inside the zone of
severe damage and four instruments were located in the sur-
rounding area west of Istanbul. Dilovasi, 35 km west of the main-
shock epicenter, is the site of planned port facilities and the
northern abutment of a planned suspension bridge across the
Gulf of Izmit. Two instruments were located in Dilovasi, one on
rock and one on articial ll. Korfez, 10 km west of the main-
shock epicenter, suffered severe damage and four instruments
were set up, three in a triangle-shaped array on young deposits
near the city center (at the TUPRAS campus), and one on older
deposits several kilometers away. Adapazari, 40 km east of the
main-shock epicenter, was also heavily damaged and three
instruments were located in a triangular array in an alluvial valley
south of the city center (at the Toyota assembly plant).
Figure 30

B

illustrates the breadth and quality of the USGS
ground-motion data. The plot shows three-component seismo-
grams recorded at selected sites during a magnitude 5.2 after-
shock (g. 30

A

) on August 31. Figure 32

A

shows three-
component seismograms (low-pass ltered at 4 Hz) recorded at
the Dilovasi and Toyota sites recorded during a magnitude 4.1
aftershock (g. 30

A

) on August 31. Plotted on a common
amplitude scale, these seismograms illustrate two site effects.
First, horizontal shaking at Dilovasi is clearly amplied at the
seismograph located on ll when compared to the shaking at
the seismograph located nearby on rock. Second, shaking is
much stronger at the two Toyota plant sites in the alluvial valley
south of Adapazari than at Dilovasi, even though the Toyota
plant sites are located farther from the epicenter of the after-
shock. Figure 32

B

shows the coherence of low-frequency
ground motion (from the magnitude-5.2 aftershock, low-pass
ltered at 0.25 Hz) recorded at each station in the triangular
array in Korfez. Figure 33

A

is a map of the Avcilar area show-
ing units of surface geology categorized according to geotech-
nical properties. Low-frequency ground motion (from the
magnitude-5.2 aftershock, low-pass ltered at 0.25 Hz) is
strongly amplied at sites in the damage area at Avcilar com-
pared to nearby sites outside the damage area (g. 33

B

).
The USGS portable instruments were placed to conrm the
importance of local geologic conditions as a mechanism for con-
centrating damage from shaking. The instruments at Avcilar,
Dilovasi, and Adapazari all revealed local amplication of
ground motion. The geologic control in Avcilar is of particular
concern because, as discussed previously, the earthquake poten-
tial south of Istanbul may have been increased by the Kocaeli
earthquake. In addition, understanding the specic geologic con-
trols that contributed to damage in Avcilar during the Kocaeli
earthquake may provide insight about what can be expected else-
where in the world. For example, there is a striking resemblance
between ground-motion and damage patterns in the Avcilar dis-
trict in Istanbul in 1999 and the Marina district in San Francisco
in 1989 (ORourke, 1992). Worldwide, many cities are situated
in similar geologic and geotechnical settings.

Figure 32.

A

, Pseudo-record section of ground motions from two pairs
of stations at Dilovasi and Adapazari plotted on the same amplitude and
time scales. Note the S-minus-P time of the first pair is approximately 5
s and that of the second is approximately 16 s.

B

, Overlay of Z, R, T com-
ponents of closely spaced North, South, West stations at Korfez, all
traces plotted on the same amplitude and time scales.

Illustration by
Edward Cranswick.

Figure 33 (facing page).

A

, Geotechnical surface map of Avcilar dis-
trict and West Istanbul. 1-3, alluvium with low-loading capacities where
load capacity increases from 1 to 3 (3 not mapped in this area); 4, alluvium
with good loading capacity with weak zones locally; 5, alluvium with low-
loading capacity with increase probability of landslides on slopes reach-
ing 14-15; 6, alluvium with high-loading capacity but landslides may oc-
cur where ground water accumulates at bottom of slopes; 7, alluvium
with low-loading capacities with known damage to structures; 13, rock
with high-loading capacities except on slopes.

B

, Low-frequency ground
motion is amplified at sites in the damage area of the Avcilar district, sta-
tions AB1, AC1, and AD1 (in red). Only radial components are displayed.

Illustration by Mark Meremonte.


36

Sea of
Mamara
Avcilar
Istanbul
0 10 KILOM ETERS
Portable digital seismograph
N
AT1 AB1 AC1 AD1 YS1 AR1 HL1
7
7
7
7
4
4
4
4
4
4
5
5
5
6
6
6
13
13
1
1
1
1
1
1
1
1
1
2
68
64
60
56
52
48
44
40
36
32
28
24
20
16
12
8
4
0
A
B

36


37

Liquefaction

Buildings in several communities, including Adapazari, Glck, and
Sapanca, sank or settled into the ground during the August 17, 1999, earth-
quake. In areas where buildings sank, surface manifestations of liquefaction,
a process by which shaking during an earthquake causes saturated sandy
materials to become uidlike, were observed. The observed surface manifes-
tations include sandy subsurface materials that vented to the surface, features
known as sand boils, and lateral shifting of the ground, a process known as
lateral spreading.
Settling of buildings was widespread in the city of Adapazari. Hundreds
of buildings were affected; in many city blocks, the majority of buildings
sank. Buildings in Adapazari locally sunk as much as 1.5 m (g. 34). The part
of Adapazari in which buildings settled is an area built on young ood-plain
deposits of the Sakarya River that has a shallow ground-water table. Although
most buildings simply sank upright into the soil, a few were also overturned,
which exposed their foundations (g. 35). The sunken buildings in many
cases were less damaged from shaking than adjacent buildings that were on
soils that did not liquefy during the earthquake. In addition, items on shelves
in many stores in the settled buildings were not even dislodged. This suggests
that once the foundation soils liqueed, seismic waves were unable to propa-
gate through the liqueed soil and shake the buildings that settled as violently
as adjacent buildings on nonliqueed soil. Sand boils were the primary sur-
face evidence of liquefaction (g. 36); these sand boils vented onto sidewalks
adjacent to buildings and into rst oor of buildings that settled. Many side-
walks around the buildings that settled were heaved upward, indicating that
liqueed material was pushed outward from beneath buildings as they sank

Figure 34.

Building in Adapazari that sank 1.5 m uniformly into the ground when the
soil beneath it liquefied and was pushed outward from beneath the building. Magni-
tudes of sinking were proportional to weight of building.

Upper photograph by Tho-
mas L. Holzer; lower photograph by National Institute of Standards and Technology.

37


38

Figure 35.

Buildings in Adapazari that either toppled or partially
overturned because soil beneath them liquefied and weakened
the foundation.

Upper photograph by Thomas L. Holzer; lower
photograph by Richard S. Olsen.

Figure. 36.

Sand boil on sidewalk in Adapazari.
Sand ejected from underground when sandy soils
beneath the ground-water table liquefied during the
earthquake.

Photograph by Mehmet elebi.


39

into the ground. These buildings sank due to soil bearing failure
associated with the loss of strength caused by liquefaction. Set-
tling also appears to be proportional to the weight of the building;
multistory buildings sank more than adjacent single-story build-
ings. Delayed building settlements were also reported for foun-
dations of silt that developed high pore pressures, but only
experienced marginal liquefaction.
Along the southeast shore of Lake Sapanca, soils also lique-
ed and the Lake Sapanca Hotel was dramatically affected. The
hotel, which was built on an old stream delta that had prograded
into Lake Sapanca, sank about 0.3 m into the ground. The
regional land surface around the hotel also lowered causing the
shoreline to submerge (gs. 37, 38). In addition, the hotel was
displaced about 0.5 m toward the lake by a lateral spread along
the shoreline of the lake. Large quantities of subsurface sand
vented to the surface and covered much of the driveway and the
landscaping around the hotel.
Minor sinking of buildings occurred along the shoreline of
Izmit Bay in Glck. A few multistory buildings sank about 0.1
m. Sand boils and lateral spreading, which caused parts of the
coastline to move bayward less than 0.5 m, were observed and
provide evidence of liquefaction along the coast. In the eastern
part of Glck, a broad regional subsidence occurred. The sub-
sidence was believed to be tectonic because it was bounded by the
normal fault described previously. The fault-bounded tectonic
subsidence caused a 4-km-long section of the coast near Glck
to submerge about 2 to 3 m.

Figure 37.

Submergence of shoreline at Lake Sapanca
and settlement of Sapanca Hotel. Moat adjacent to wall
indicates the hotel has sunk into the ground because of liq-
uefaction of underlying sandy soil.

Photographs by Thomas
L. Holzer.

39


40

GRAIN SIZE, IN mm
0.001 0.01 0.1 1
P
E
R
C
E
N
T

F
I
N
E
R

B
Y

W
E
I
G
H
T
0
25
50
75
100
Glck
Sapanca
Adapazari

Sand boils were sampled at Adapazari, Glck, and Sapa-
nca. Liqueed material was predominantly ne grained; grain-
size distributions of the material typically ranged from silts to
silty sands (g. 39). The two coarser sand boils sampled at the
Sapanca site contained cobbles and other coarse-grained mate-
rial. These sand boils may have been contaminated with cobbles
contained in the base of a driveway. The other sand boil at Sapa-
nca was in an open area and not susceptible to contamination.
Sunken and overturned buildings have been observed in
other large, greater than 7, magnitude earthquakes, including the
1964 Niigata, Japan, and 1990 Philippines earthquakes. Magni-
tudes of sinking of buildings during smaller, less than 7, magni-
tude earthquakes typically have been small. In general, the
duration of strong shaking is proportional to earthquake magni-
tude. These observations suggest that the magnitude of sinking is
dependent not only on the weight of the building but on the dura-
tion of strong shaking, which both keeps the liqueed soil in a
uid state and potentially liquees a greater thickness of soil.

Figure 38.

Sunken entryway into Sapanca Hotel created when building sank into liquefied soil.

Photograph by Thomas L. Holzer.

Figure 39.

Grain-sized distributions of sand boils sampled near sites
of sunken buildings in Adapazari, Glck, and Sapanca. Vertical axis is
percent of sample by weight finer than the grain size shown on the hor-
izontal axis.

Illustration prepared by Michael J. Bennett. Data from
USGS and ERDC.


41

Performance of the
Built Environment
Trans European Motorway
(TEM)
Earth Dams
Buildings

The region hit by the Kocaeli earthquake is the industrial heartland and
one of the most densely populated sections of Turkey. Ofcial government
estimates of damaged living and business units (as of October 19, 1999) indi-
cate the magnitude of the human disaster. More than 77,000 living units were
heavily damaged, more than 77,000 were moderately damaged, and about
90,000 were lightly damaged. All of the heavily damaged and most of the
moderately damaged living and business units are in buildings that are
beyond repair and will probably be demolished. The region with heavy dam-
age extends from Yalova, which is 50 km west of the epicenter, to the town of
Dzce, which is 100 km east of the epicenter. The majority of the building
collapses and deaths occurred in small towns located on the narrow at shore-
lines of the Sea of Marmara. There was also heavy damage and loss of life in
Adapazari, which is located on the ood plain of the Sakarya River. A small
western suburb of Istanbul, Avcilar, located near Istanbul Atatrk airport, suf-
fered about 1000 fatalities from collapsed buildings despite the unusually
large distance (about 80 km) from the seismic source zone.
Most residential construction in the affected region of Turkey consists
of reinforced concrete frames (column-slab systems) with masonry inll.
Multistory residential buildings, as well as buildings containing commercial
space on the rst oor and residential units above, use this type of construc-
tion. The reinforced concrete frame buildings that suffered severe structural
damage or complete collapse were typically four to eight stories in height.
Many of the collapsed buildings appeared to be relatively new and indeed
some were either under construction or recently completed, but not yet occu-
pied at the time of the earthquake.
Figure 40 shows a building under construction illustrating common con-
struction practice. In most cases of building collapse, there was no evidence
of shear walls. Consequently, lateral resistance to the strong ground shaking
was provided solely by frame action. Concrete compressive strength was not
measured, but honeycombing was observed in similar new construction that
did not collapse suggesting somewhat poorer concrete placement practices
than typically found in U.S. construction. Furthermore, undeformed reinforc-
ing bars were widely used for both exural and shear reinforcement, a prac-
tice not permitted under U.S. building codes. Most of the reinforced concrete
buildings can be characterized as having little or no ductility.

41


42

The most common type of structural failure observed
throughout the affected regions was shear failure of rst-story
columns. In many instances, the rst story collapsed while the
upper stories remained relatively undamaged as seen in the 5-
story apartment building in Yalova shown in gure 41. In other
instances, shear failure of intermediate columns triggered partial
collapse of the buildings as illustrated in gure 42. Hinging of
rst-story columns was also observed (g. 43).
In Adapazari, part of which is situated on soils that liqueed
during the ground shaking, many buildings sank due to soil fail-
ure as previously discussed. Figure 34 shows several commercial
buildings that sank and heaved the surrounding brick sidewalk.
In other instances, soil failure contributed to overturning of
buildings with narrow footprints as seen in gure 35.

Figure 40.

Building in Izmit under construction using reinforced con-
crete frame construction. Note absence of shear walls.

Photograph by
National Institute of Standards and Technology.

Figure 41.

Partial building collapse in Yalova as the result of shear fail-
ure of first story columns.

Photograph by National Institute of Standards
and Technology.

Figure 42.

Partial collapse of building near Izmit as the result of shear
failure of intermediate columns. Note various stages of damage despite
apparently similar construction.

Photograph by National Institute of
Standards and Technology.


43

Figure 43.

Near collapse of building in Adapazari as
the result of formation of column hinges in the first
story.

Photograph by National Institute of Standards
and Technology.

Figure 44.

Damage to Officers Club on Turkish Naval Base, Glck,
which was located directly on the surface fault rupture.

Photograph by
National Institute of Standards and Technology.

Figure 45.

Industrial building damaged by near-fault ground move-
ments east of Glck. Building is on downthrown block of normal fault
and near a major change of strike in the surface fault scarp.

Photograph
by National Institute of Standards and Technology.


44

Zone 1
Zone 2
Zone 3
Zone 4
Zone 5
Seismic zones
0 100 KILOM ETERS
N
Active fault

Locally, surface fault rupture caused collapse of or damage
to many structures. The majority of structures directly atop the
rupture either collapsed or were seriously damaged. One exam-
ple of fault damage is the Ofcers Club located on the Turkish
Navy Base at Glck (g. 44). Lateral displacement of more
than 2 m was observed in the portion of the building directly
over the fault resulting in signicant damage, while adjacent sec-
tions of the building fared better exhibiting less damage.
Differential ground movement adjacent to fault rupture was also
responsible for building damage in a few cases as seen in gure
45. An industrial building under construction was deformed by
approximately 0.5 m of differential settling associated with a
change in strike of the trace of a normal fault that ruptured dur-
ing the Kocaeli earthquake.
Turkey has adopted modern building codes with up-to-date
seismic design provisions. According to the Turkish seismic zone
map, the areas affected by the Kocaeli earthquake are in the zone
of highest risk (g. 46). Although only examples of severe dam-
age and collapse have been reported here, many buildings in the
affected areas performed well with little or no apparent damage.
Indeed, on gure 42, three buildings, one of which collapsed
completely, one of which suffered partial collapse, and one of
which appears to be relatively undamaged are visible. The
absence of shear walls, decient detailing around beam-column
joints, and use of smooth rather than deformed steel reinforce-
ment may have contributed to the extent of damage and collapse
by reducing the ductility of the buildings. However, the heavy
damage or collapse of some buildings where nearby buildings of
similar size, conguration, and construction survived relatively
intact, suggests that the poor performance of some buildings may
be traced to failure to design or construct in compliance with
established building codes. The contrasting performance
between similar buildings that survived and those that collapsed
or were heavily damaged illustrates that modern building codes,
when complied with, can help prevent or limit disaster.
The Kocaeli earthquake also demonstrated the hazard posed
by aftershocks to buildings following a large earthquake. Figure
47 shows a building in Glck before and after it collapsed dur-
ing a M

W

5.9 aftershock on September 13. This aftershock,
which was the largest magnitude aftershock at that time to follow
the August 17 earthquake, killed 7 people, injured at least 239
people, and caused dozens of buildings to collapse in three cities
near the epicenter.

44

Figure 46.

Modified earthquake zoning map of Turkey.

Modified from Earthquake Research Department, Directorate for Disaster Affairs, Ministry
of Public Works and Settlement.

Figure 47 (pages 4546).

Multistory residential building in Glck. Tec-
tonic subsidence associated with faulting in the main shock caused the
inundation of the first floor of the building. Upper photograph shows
building standing after the August 17 main shock and before it collapsed
in an aftershock;

photograph by Thomas L. Holzer

. Lower photograph
shows building collapsed by M

W

5.9 aftershock four weeks after the
main shock;

copyrighted photograph by AP/Wide World Photos, used
with permission.

45 46

47

Performance of the
Built Environment
(TEM)
Earth Dams
Buildings

In 1985, Turkey initiated an ambitious national motorways program
along its most heavily used transport corridors. This program has cost more
than $13 billion and included planned construction of 1,900 km of motorway.
Approximately 1,600 km of motorway have been completed. The Trans Euro-
pean Motorway (TEM) is an important part of the motorway system, which
connects Asia to Europe. The TEM runs westward from Ankara, the capital
city of Turkey, to Turkeys border with Bulgaria and Greece, where it connects
to other European highways.
The TEM was constructed following a modern construction code similar
to the American Association of State Highway and Transportation Ofcials
(AASHTO) standard. As such, its performance during the Kocaeli earthquake
may provide some insight into how American highway systems will perform
in similar situations.
Immediately following the August 17, 1999, earthquake, the section of
the TEM east of Izmit was closed because large cracks and areas of settlement
on the roadway, one collapsed overpass, and the need to inspect all of the
bridges. Temporary repairs were made and the TEM was reopened to the traf-
c after three days. In some areas, however, trafc was limited to two-way
trafc on one side and speed limits were imposed while repairs continued (g.
48). Most repair work was completed within 18 days following the earthquake
and the TEM returned to normal operation.
Statistics for damage sustained along the TEM between Camlica, Istan-
bul and Gumusova, Bolu are summarized in table 2. Although a signicant
portion of the TEM is in the area strongly shaken by the Kocaeli earthquake,
damage to bridges along the motorway was restricted to the section between
5 km west of Izmit and the Sakarya River. Densication of dry roadway lls
and bridge and culvert approaches locally caused settlements of the roadway
surface in the section extending from about Izmit to 50 km east of Izmit. At
several locations, surface faulting along the North Anatolian fault offset the
TEM (g. 49). The TEM runs parallel to the North Anatolian Fault on the
south side of Lake Sapanca.

Figure 48.

Following the earthquake, road surface repairs forced alternative use of the eastbound and west-
bound lanes of the Trans European Motorway as a two-way highway with greatly reduced speeds from Izmit to
about 50 km east of Izmit.


47


48

Strong ground shaking caused slight to moderate damage to
bridges and overpasses. The most common types of damage were
displacements of girders at seatings, shear key failures, and cover
concrete spalling of deck around abutments. For example, at the
Cinarlidere viaduct both shear key failure and cover concrete
spalling of deck around an elastomeric bearing pad are visible
(g. 50). Additional instances of cover concrete spallings of deck
were observed as a result of the deck pounding the abutment.
At the Sakarya River viaduct, bearings were unseated and
the bridge carrying the westbound lanes of the TEM over the
Sakarya River had to be
closed to trafc for repairs.
The damage was more sig-
nicant at the east abut-
ment side of the bridge.
Approximately 50 cm of
transverse displacement of
the span with respect to the
east abutment is visible on
gure 51. This was the
largest permanent displacement caused by shaking at the TEM
bridges. As a result of strong ground shaking, shear keys failed
and the span was unseated. The adjacent bridge carrying the east-
bound lanes of the TEM sustained less damage and remained
fully operational.

Figure 49.

The North Anatolian fault, which crossed the Trans Europe-
an Motorway at several locations in Turkey, caused major disruptions to
the roadway surface.

Photograph by Robert K. Langridge.

Figure 50.

Failure of shear keys was common at bridges of the Trans
European Motorway. Shear keys at the end of girders prevent transverse
and longitudinal seismic movements.

Photograph by Selcuk Toprak.

Motorway feature Total Damaged

Roadway 183 km 50 km
Tunnel 4 ---
Viaduct 28 4
Overpass, underpass,
and stream bridges
143 8
Overpass, underpass,
and stream bridge approaches
(differential settlement)
143 26

Table 2.

Statistics for earthquake damage along Trans European
Motorway between Camlica and Gumusova, Turkey.

Concrete spalling:

Pieces breaking off the
surface of main concrete structure.

Shear key:

Concrete structure at abut-
ments that inhibit transverse and longitu-
dinal seismic movements of girders.

Soil densication:

Decrease in volume of
a block of granular material.


49

Figure 51.

Ground shaking caused the
westbound span of the Sakarya River bridge
to displace 50 cm relative to the east abut-
ment. This was the largest permanent dis-
placement caused by shaking at any bridge.

Photograph by Selcuk Toprak.

Figure 52.

Densification of dry ap-
proach fill adjacent to bridges of the Trans
European Motorway was widespread and
caused disruption of the roadway. Imme-
diately following the earthquake, Turkish
highway engineers placed asphalt ramps
on the roadway to reduce the hazard.

Photographs by Thomas L. Holzer.


50

Permanent ground displacements, such as extensive cracks
and settlement, caused signicant damage to the roadway sur-
face. Approximately 50 km of the TEM required resurfacing.
Densication of dry ll commonly caused settlements in areas
where the motorway approached bridges, viaducts, and culverts;
some areas settled as much as 0.3 m. Figure 52 shows one exam-
ple of settlement.
The surface fault crossed the TEM at several locations and
caused the collapse of the Ariye overpass (g. 53). Field obser-
vations revealed that the fault rupture passed directly beneath the
north end of the overpass. Fault displacement along the fault near
the overpass was 3.60.3 m. This large fault displacement caused
the spans to fall to the ground. Immediately after the collapse, a
bus crashed into one of the fallen spans killing 10 people.

Figure 53.

The Arifiye overpass collapsed because the surface fault rupture passed directly through its north abutment (foreground). Ten riders of
a bus were killed when the bus collided with the collapsed overpass.


50


51

Performance of the
Built Environment
(TEM)
Earth Dams
Buildings

Four major dams within 45 km of the fault rupture were visited by the
team from the Engineer Research and Development Center of the U.S. Army
Corps of Engineers (g. 54). Two dams north of the Izmit Bay, Omerli Dam
and Darlik Dam, were in the area where ground motions were inferred to be
less than 0.2 g. Two other major dams south of Izmit Bay, Kirazdere
(Yuvacik) Dam and Gokce Dam, were in the area where peak ground accel-
erations were inferred to be equal to or greater than 0.2 g. Kirazdere Dam is
about 4 km from the causative fault. The dam may have experienced peak
horizontal motions of 0.3 to 0.4 g. Gokce Dam is in the area of considerable
damage and numerous aftershocks in the area of Yalova. Gokce Dam may
have experienced motions of at least 0.2 g or greater (table 3).
Omerli Dam holds the primary water reservoir supplying the Asian side
of Istanbul, that part of the city that lies on the east side of the Bosphorus. The
soil and rock dam is of critical importance for Istanbul. The reservoir at the
time of the inspection was 5 m below the spillway invert. The upstream face
of the dam is composed of rocks as large as 1 m in diameter. The downstream
area of the dam is covered with grass with shrubs as high as 1 m. No cracking
was observed during the site visit or by the guard who was at the dam the
morning after the earthquake. There may have been water seepage at the
downstream toe because the observed type of vegetation requires lots of
water. No seepage or effects of seepage were seen at the junctions of the
embankment to the valley wall. The upstream slope down to the reservoir
level is not uniform. Specically, just above the reservoir level, the upstream
slope had a bump. The non-uniform bump had an approximate height of
2+ m and a length along the slope of 5 m. It is not known if the feature was
caused by shallow slumping during the earthquake.
Darlik Dam is composed of a rock-lled shell with a clay core. The
upstream and downstream faces are composed of 1-m-diameter angular rock.
The reservoir level during the earthquake appears to be at 60 percent capacity
relative to the spillway invert. The access to the crest of the dam was from a
road on the downstream slope. No damage or deformation was observed at
Darlik Dam. No cracks on either the asphalt crest road or adjacent gravel side
were observed during the site visit or reported by the manager of the dam. The
upstream and downstream face of the dam appeared to be very uniform when
observed from the crest and downstream toe. No movement or cracks were
observed on the access road. There was also no evidence of recent water seep-
age or sand boils near the downstream toe or sidewalls of the dam.

51


52

Omerli Dam
Darlik Dam
Gokce Dam
Kirazdere (Yuvacik) Dam
BLACK SEA
SEA OF MARMARA
ISTANBUL
Izmit
Adapazari
N
EPICENTER
41
41
30'
30
31
29
0 50 KILOM ETERS

52

Figure 54.

Locations of earth dams visited by U.S. Army Corps of Engineers representatives.

Base map courtesy of BKS Surveys Ltd., N. Ireland.

Earth dam Height from
foundation,
in m
Height from
river bed,
in m
Crest
length,
in m
Reservoir
area,
in km

2

Reservoir
volume,
in km

3

Distance
from fault,
in km

Omerli 54 52 372 23 0.386 43
Darlik 47 45 308 5.56 0.107 45
Gokce 47 50 748 1.28 0.013 36
Kirazdere
(Yuvacik)
108 102 390 1.74 0.005 4

Table 3.

Descriptions of major earth dams in the epicentral area of the Kocaeli, Turkey,
earthquake.


53

The Gokce Dam is composed of a rock-lled shell and a
clay core with upstream and downstream sand lters and is
founded on bedrock. The reservoir level during the earthquake
appears to be at 70 percent in terms of the spillway invert.
Gokce Dam is estimated to have experienced 0.2 g
or greater of peak horizontal acceleration. The only
observed effect of the earthquake on the dam was a
longitudinal crack that developed along the full
length of the crest. The crack was located along the
upstream side of the gravel road that runs along the
crest of the dam. The maximum width of the crack was 7 mm at
the middle of the dam and the upstream side of the crack
appears to have slightly dropped down. Fine cracks, not affect-
ing the structure, were observed in a concrete pump station
building at the dam.
The downstream face of the dam was observed to be some-
what irregular; the irregularity is on the order of 1 to 2 m. There
was no evidence that this irregularity was the result of localized
slumping and there were no indications of fresh soil movement
such as effects on plants or exposures of soil surfaces. The
exposed upstream face appears to have a more uniform appear-
ance. There was also no evidence of recent water seepage or sand
boils from piping near the downstream toe of the dam. Except for
the longitudinal crack at the crest, the dam showed no effects of
the earthquake shaking.

Figure 55.

View of Kirazdere (Yuvacik) Dam after the Kocaeli earthquake. The dam was only about 4 km from the surface rupture and did not suffer
damage.

Photograph by Richard S. Olsen.

Spillway invert:

The lowest point on the
passage in or about a dam for the escape
of surplus water.


54

Kirazdere Dam, also known as Yuvacik Dam, is an earth-
ll dam with a clay core, sand and gravel lters, and a rock shell
(g. 55). Kirazdere Dam is of particular interest because it is
only 4 km from the surface rupture associated with the Kocaeli
earthquake. The dam is inferred to have experienced 0.3 to 0.4 g
of peak horizontal acceleration, a shaking duration of 30 s with
accelerations greater than 0.05 g, and a spectral range of excita-
tions from 0.2 to 6.0 Hz. However, the dam appears to have suf-
fered no damage from the earthquake. The dam manager stated
that a few short 2-mm-wide cracks resulted from the earthquake
on the gravel road along the crest near the left abutment looking
upstream, parallel to the axis of the dam. The lengths of these
cracks were not measured. The road on the crest is heavily used
and the cracks quickly disappeared. A few hairline cracks were
seen in a one-story concrete structure at the dam. No cracks
were seen in the concrete of the spillway.
Benches excavated into the left abutment, looking up-
stream, showed some material that had fallen away from the
stepped benches and small accumulations of talus and rockfall
debris. The dam manager explained that although some of this
material fell during the earthquake, some was there before the
earthquake. The cracks and talus on the abutment were
extremely minor and the upstream and downstream faces of the

54

dam were straight and uniform. An underground water supply
pipeline from the dam was damaged during the earthquake and
was being repaired during the eld visit.
The notable observation about earth dam performance dur-
ing the Kocaeli earthquake is the absence of damage. No dams
exhibited any damage other than cracks that opened a few milli-
meters. Two earth dams on bedrock that were examined experi-
enced earthquake shaking with peak accelerations greater than
0.2 g; these were Gokce Dam where the peak acceleration was
approximately 0.2 g, and Kirazdere (Yuvacik) Dam where the
peak acceleration was as much as 0.4 g. The behavior of
Kirazdere Dam, which is located practically at the epicenter of
the Kocaeli earthquake, is of particular interest because the dam
experienced a very signicant (M

W

7.4) earthquake with no
damage. The spectral range of the excitations was extremely
broad, from 0.2 to 6.0 Hz. The duration of shaking with acceler-
ations greater than 0.05 g was around 30 s. Gokce Dam is situ-
ated in an area of considerable damage and numerous
aftershocks and may have experienced motions of at least 0.2 g
or greater. The only effect of the earthquake at Gokce Dam was
an explainable minor crack along the crest.


55

Implications for the
United States
San
Francisco
Los Angeles
Izmit
TURKEY
CALI FORNI A

Earthquakes in urban areas offer special opportunities to evaluate the
effect of shaking and secondary effects from earthquakes on the built environ-
ment. Although on average, about 19 earthquakes with magnitude greater
than 7 occur annually worldwide, few of these occur in or near urban areas
where substantial damage occurs to buildings and other civil works. The M

W

7.4 Kocaeli, Turkey, earthquake is one of these few earthquakes. Large earth-
quakes impact almost all elements of the urban environment and impact large
numbers of people making it challenging to single out the most important les-
sons to be learned. However, some important implications for the United
States can be drawn from the Kocaeli earthquake and they are outlined below.

LARGE-SCALE HUMAN DISASTER

Of all natural disasters, large magnitude earthquakes in urban areas are
almost unique in their ability to cause large-scale human suffering and prop-
erty loss. Only hurricanes and typhoons can compare. That the Kocaeli earth-
quake damaged homes and displaced a quarter of a million people illustrates
the scale of the loss (g. 56). Urban areas in the United States in which poten-
tial seismic sources have been recognized include Salt Lake City, Seattle, San
Francisco Bay region, greater Los Angeles area, and greater San Diego area.
Computer models, such as the Federal Emergency Management Agency
HAZUS earthquake loss model (National Institute of Building Sciences,
1997), predict that similar tragedies are possible here in the United States. For
example, these models predict that a magnitude 7.2 quake on the San Fran-
cisco Peninsula would displace more than 100,000 people from their homes
while a magnitude 7.3 on the Hayward fault, which passes through Oakland,
Calif., would displace about 150,000 people. A magnitude 6.6 on the northern
part of the Hayward fault would probably displace more than 30,000 people.

BUILDING COLLAPSE HAZARD

The large number of collapsed buildings in Turkey was the principal
cause of the large death toll in the Kocaeli earthquake. More than 20,000
buildings are estimated to have collapsed. Nearly all of these buildings, in
addition to the ones that were damaged beyond repair, are made of reinforced
concrete, a common construction style both in the United States and world-
wide (g. 57). Although modern engineering design of reinforced concrete
buildings, particularly in areas of high seismic risk, incorporates ductility that
permits new buildings to withstand the oscillatory motions imposed by earth-
quakes, much of the building stock in the United States was constructed
before the importance of ductility was fully appreciated. In the Eastern United
States most reinforced concrete structures, including both buildings and
bridges, have little or no ductility. This leaves the United States vulnerable to
the catastrophic collapse of large numbers of buildings and bridges when they
are shaken by the oscillatory motion of large earthquakes. The collapse of the
Cypress Street viaduct in Oakland, which killed 42 people during the 1989
Loma Prieta, Calif., earthquake, is a recent example in the United States of the
seismic failure of a concrete structure with little or no ductility.

55


56

EARTHQUAKE MONITORING SYSTEMS

A belated reaction by government emergency response of-
cials in Turkey to the earthquake disaster may have been due in
part to the inability of the local seismic monitoring network to
determine a reliable seismic magnitude and shaking intensities
for the Kocaeli earthquake. Near real-time warnings using mod-
ern digital technology are now available and can identify pre-
cisely both the size and location of a large earthquake within
minutes of its occurrence as well as the distribution of shaking
intensities. This was most recently demonstrated in Taiwan,
where governmental ofcials were alerted within two minutes of
the occurrence of the M

W

7.6 Chi-Chi earthquake of September
21, 1999. Most regional seismic networks in the United States
rely on old analog technology and await modernization.

Figure 56.

On November 16, 1999, President Clinton greets a displaced Turkish family living in a temporary tent city near Izmit since the August 17
Kocaeli earthquake.

Copyrighted photograph by J. Scott Applewhite, AP/Wide World Photos, used with permission.

57 58
Figure 57. Street scene in Adapazari following the Kocaeli, Turkey, earthquake. Extensive damage to concrete structures with little or no ductility
displaced more that a quarter of a million people from their homes in the epicentral region. Photograph by Thomas L. Holzer.

59

EARTHQUAKE FORECASTING

The details of how the ground ruptured along the North
Anatolian fault will be useful in the understanding of how faults
in the United States may rupture in future earthquakes. Of par-
ticular interest is where and how the surface ruptures started
and stopped in relation to geometric changes in the North Ana-
tolian fault, the amount of slip, and the length of rupture. Field
studies of surface ruptures from major earthquakes permit geo-
scientists to better understand the faulting process and the char-
acteristics of segmented faults. Identication of fault segments
likely to rupture in an earthquake is an important step in the
forecasting of both magnitude and location of future earth-
quakes in the United States.
The North Anatolian fault system is a close analog of the
San Andreas fault system in California. Both are boundaries
between tectonic plates and have similar slip rates, total length
and straightness (g. 58). The systematic westward progression
of earthquakes along 1000 km of the North Anatolian fault since
1939 suggests that each of the earthquakes in the progression in
Turkey has triggered the next one. It is likely that the November
12, 1999, M

W

7.1 earthquake near Dzce was triggered because
it occurred in an area where stress was increased by the August
17, 1999, main shock. Understanding this process may be useful
for improving probabilistic earthquake forecasts in the United
States. Such forecasting is increasingly being applied to the San
Andreas fault system, where recent forecasts indicate a 70 per-
cent probability of an M

w

>6.7 in the San Francisco Bay region
in the next 30 years. The probability of an earthquake with a
magnitude greater than or equal to that of the M

w

7.4 Kocaeli
earthquake is 13 percent.

PERMANENT GROUND DEFORMATION HAZARDS

While earthquake shaking typically causes more than 90
percent of the loss associated with earthquakes, permanent defor-
mation of the ground locally can be more devastating to struc-
tures. Few structures are built to withstand large permanent
ground movement. Three causes of permanent ground deforma-
tion were noteworthy in the Kocaeli earthquake: liquefaction,
surface faulting, and densication of dry soil.
The buildings that sunk in Adapazari and other communities
because of liquefaction demonstrate one effect of long duration
shaking associated with larger magnitude earthquakes. Numer-
ous multistory buildings in Adapazari suffered the dramatic con-
sequence of foundation liquefaction and the resulting loss of
strength. These buildings experienced massive foundation-bear-
ing failures resulting in buildings either sinking vertically into the
ground or overturning.
Liquefaction of silts observed in the city of Adapazari has
important implications for earth dams founded on alluvial foun-
dations around the world. Low-strength silt and soil mixture
foundations beneath earth dams can liquefy and cause the
embankment to sink or rotate into the foundation. The conse-
quence could be large deep cracks near or through the clay core
with the potential for release of the reservoir. Historically, silts
have been observed to liquefy during earthquakes, however,
many believed that the consequences would be less severe than
where clean sands liqueed. The performance of buildings
founded on weak silts in Adapazari during the Kocaeli earth-
quake has advanced our understanding of liquefaction effects
and has shown how earth dams founded on weak silts could
deform and fail. Settlement during the earthquake of properly
compacted earth dams founded on rock foundations in Turkey,
however, was minimal.
The bearing capacity failures associated with the sunken
buildings may also be relevant to the response of levee systems
in the United States when they are shaken by earthquakes.
Levees typically are built in ood plains, where the soils are
highly susceptible to liquefaction when shaken strongly. Levee
systems along the Mississippi River in the New Madrid seismic
region of the Central United States are potentially vulnerable.
Liquefaction during a repeat of the 1811-12 earthquake
sequence could result in widespread lowering of levees along
the entire Mississippi River system. The stability of the levee
system in the Sacramento-San Joaquin River delta, California is
likewise threatened. This hazard compounds the existing chal-
lenge of dealing with the extensive potential for static failures of
levee systems in the United States.


60

San
Francisco
Los Angeles
Salton
Istanbul
Erzincan
TURKEY
CALI FORNI A
Fault slip rate
24
+
4 mm/yr
_
Fault slip rate
2034 mm/yr
Izmit
EPICENTER
19391999 Earthquake Sequence
19391944 Progression
1857 Earthquake
1906 Earthquake

The 110-km-long surface fault rupture associated with the
earthquake was particularly damaging because it intersected
many developed regions. Although California has implemented
an effective program through the Alquist-Priolo Act, which was
enacted after the 1971 San Fernando, Calif., earthquake, to pre-
vent signicant new construction on active surface faults, older
construction and lifelines continue to be at risk. Other parts of the
country with mapped active faults have yet to regulate develop-
ment on fault lines. For example, on the western front of the
Wasatch Mountains in Utah, development on well known active
faults is left to the option of and regulated by individual cities,
which has left some populated areas at risk.
The Turkish segment of the Trans European Motorway
(TEM), which was built to the standards of the American
Association of State Highway and Transportation Ofcials
(AASHTO), was seriously disrupted by permanent ground
deformation. The kinds of deformation that affected the TEM,
which included both faulting and densication of dry soils, are
not addressed in the AASHTO code. Operation of the TEM
was impacted when approaches to bridges that settled required
extensive resurfacing and at one location when surface fault-
ing through an overpass caused its collapse and temporarily
halted trafc during the critical period just after the earth-
quake. The implications to the highway system in the United
States are profound.

60

Figure 58.

The striking resemblance of the San Andreas and North Anatolian fault systems is visible in this comparison. This suggests that under-
standing of fault mechanics from one fault system may be applied directly to the other.

Illustration prepared by Ross S. Stein.


61

Acknowledgments

Kandilli Observatory and Earthquake Research Institute (KOERI) of
Bogazici University in Istanbul, Turkey, was the USGS host and collaborator
for the post-earthquake investigations and aftershock monitoring. We thank
Professor Ahmet Mete Isikara, the Director of the Observatory, and Professor
Mustafa Erdik, the Head of the Institute, for their hospitality and help, and all
the researchers and graduate students who participated in our investigations,
instrument deployment, and data analysis.
The USGS fault rupture investigation was coordinated through Istanbul
Technical University. It received generous guidance and assistance from Pro-
fessor Aykut Barka as well as Professors Serder Akyuz and Erhan Altunel of
Osmangazi University.
Many people generously assisted the investigation of earth dams by the
Engineer Research and Development Center of the U.S. Army Corps of Engi-
neers. Professor Mustafa Erdik furnished strong motion instrumental data and
background for understanding the earthquake processes in the region. He also
provided contacts with Turkish ofcials for permission to visit dams. Assis-
tance in making contacts was also provided by Ms. Juana Alsace and
Ms. Mary Ann Whitten of the U.S. Consulate in Istanbul. The Operations
Disaster Center at the Istanbul International Airport where many Turkish gov-
ernmental agencies were represented was visited. Major Bill Pitts,
U.S. Army, put the group in touch with Dr. Tanay Sidki Vyar, Technical
Advisor for State Affairs of Water, Turkey, and Mr. Mehmet Duman, Istanbul
Director of Foreign Relations. Meetings were arranged with Mr. Halil Tanir,
head of the Water Construction Department, Istanbul Water Administration
(ISKI), Mr. Cemaletti Yicher, geologist with ISKI, and Mr. Ahmet Otbeli,
civil engineer with ISKI. Further meetings were held with Mr. Lut Aydin,
Director of Administration, ISKI, and Dr. Veysel Eroglu, Director General of
the Istanbul Water Administration. Field visits were arranged for the dams.
The investigation of structural damage by the National Institute of
Standards and Technology was assisted by the reconnaissance team of the
Earthquake Engineering and Research Institute. Many individuals shared
information and provided transportation and access to sites of damage in
Turkey. They included Professor T. Leslie Youd (Brigham Young Univer-
sity), Professor Mark Ascheim (University of Illinois at Urbana/Cham-
paign), Professor John Mander (SUNY at Buffalo), Professor Mete Sozen
(Purdue University), Gayle Johnson and Peter Yanev (EQE International,
Inc.), Professor Bill Mitchell (Baylor University), and Professor Emre Otay
(Bogazici University).

61


62

The following people at Directorate of Highways, District 1,
provided highway damage information and escorted U.S.
researchers to Trans European Motorway bridge sites: Yalcin
Yumrukcalli, Tarcan Atay, Dogan Akgun, Yavuz Oktay, Mesut
Korukluoglu, and Unal Anakok. We especially thank Yavuz
Oktay for excellent feedback to follow-up questions. Professor
Kutay Ozaydin of Yildiz Technical University arranged the meet-
ing at Directorate of Highways, District 1.
We gratefully acknowledge the following people who pro-
vided either information or assistance in the eld: Gulum Bir-
goren, Muzaffer Gul, Mehmet Cem Ozbey, Kemal Beyen, Seref
Polat, Hafez Keypour (all from KOERI), Mohammad R. Zolf-
aghari (Applied Insurance Research, Ltd.), Gerzel Sunal, Aynur
Dikbas, Zya Cakir, M. Kaya, B. Yerli (all from Istanbul Technical
University), Professor James Dolan, Ross Hartleb, Shari Christ-
offerson, Alan Tucker (all from University of Southern California
and Southern California Earthquake Center), Dr. R. Armijo, Dr.
B. Meyer, Dr. J. B. de Chabalier (all from Institut de Physique du
Globe de Paris), Professor Atilla Ansal (Istanbul Technical Uni-
versity), Professor Polat Gulkan (Middle East Technical Univer-
sity, Ankara), Mr. Engin Inan and Mr. Huseyin Guler (Earthquake
Research Department, Directorate for Disaster affairs, Ministry
of Public Works and Settlement), Mr. Ruchan Yilmaz (Director
for Directorate for Disaster affairs, Ministry of Public Works and
Settlement), Oktay Ergunay (Ministry of Public Works and Set-
tlement), Professor Dogan Altinbilek (Director General, Public
Water Works, Ankara), and Professor Cetin Yilmaz (Middle East
Technical University, Ankara).
Robert D. Brown, Jr., and James W. Dewey reviewed
early drafts of the circular and made many constructive sug-
gestions. Michael J. Bennett, Joel Robinson, Selcuk Toprak,
and Nancy Shock created drafts of illustrations. The base map
used in gures 22 and 54 was provided courtesy of BKS Sur-
veys Ltd., N. Ireland.
The Circular was edited by S.J. Kropschot and coordination
of graphics, layout, and design was by Carol A. Quesenberry.


63

References

Ambraseys, N.N., 1970, Some characteristic features of the Anatolian fault zone:
Tectonophysics, v. 9, p. 143165.
Ambraseys, N.N., and Finkel, C.F., 1995, The seismicity of Turkey and adjacent areas,
A historical review, 15001800: Istanbul, Muhittin Salih Eren, 240 p.
Ambraseys, N.N., and Zatopec, A., 1969, The Mudurnu Valley, West Anatolia, Turkey,
Earthquake of 22 July 1967: Seismological Society of America Bulletin v. 59, no. 2,
p. 521589.
Armijo, R., Meyer, B., Hubert, A., and Barka, A., 1999, Westward propagation of the
North Anatolian Fault into the northern AegeanTiming and kinematics: Geol-
ogy, v. 27, p. 267270.
Aydinoglu, N., trans., 1998, Specification for structures to be built in disaster areas:
Ministry of Public Works and Settlement Government of Republic of Turkey,
Issued on 2.9.1997, Official Gazette no.23098, effective from 1.1.1998; amended on
2.7.1998, Official Gazette no.23390.
Barka, A., 1992, The North Anatolian fault zone: Annales Tectonicae, v. 6, suppl.,
p. 164195.
Barka, A., 1996, Neotectonics of the Marmara regionSelected papers on the North
Anatolian fault, field trip to the North Anatolian fault, August 115, 1997: Istanbul,
Istanbul Technical University and Eurasia Earth Science Institute, 56 p.
Barka, A., 1999, The 17 August 1999 Izmit earthquake: Science, v. 285, p. 18581859.
Barka, A., Akyuz, S., Altunel, E., Sunal, G., Cakir, Z., Dikbas, A., Yerli, B., Rockwell, T.,
Dolan, J., Dawson, T., Hartleb, R., Tucker, A., Fumal, T., Langridge, R., Stenner, H.,
Christofferson, S., Armijo, R., Meyer, B., and Chabalier, J., 1999, 17 August 1999
Izmit Earthquake, Northwestern Turkey [abs.]: Eos, in press.
Boore, D.M., Joyner, W.B., and Fumal, T.E., 1994, Estimation of response spectra and
peak accelerations from western North American earthquakesAn interim
report; Part 2: U.S. Geological Survey Open-File Report 94127, 40 p.
Federal Emergency Management Agency, 1998, 1997 edition, NEHRP recommended
provisions for seismic regulations for new buildings and other structures, Part 1
Provisions: Washington, D.C., Building Seismic Safety Council, FEMA 302, 337 p.
Fumal, T.E., Langridge, R.M., Stenner, H.D., Christofferson, S., Dolan, J.F., Hartleb, R.,
Tucker, A.Z., Rockwell, T., Dawson, T., akir, Z., Dikbas, A., Yerli, B., and Barka, A.,
1999, Slip distribution and geometry of the Sakarya section of the 1999 Izmit earth-
quake ground rupture, Western Turkey [abs.]: Eos, in press.

Hartleb, R., Dawson, T., Tucker, A., Dolan, J., T., Rockwell, J., Yerli, B., Dikbas, A., akir,
Z., Gurer, T., Uslu O., and Barka, A., 1999, Surface rupture and slip distribution
along the Dzce strand of the 17-August-1999 Izmit, Turkey earthquake [abs.]: Eos,

in press.
Ikeda, Y., Suzuki, Y., Herece, E., Saroglu, F., Isikara, A.M., and Honkura, Y., 1991, Geo-
logical evidence for the last two faulting events on the North Anatolian fault zone
in the Mudurnu Valley, western Turkey: Tectonophysics, v. 193, p. 335345.
Nalbant, S.S., Hubert, A. and King, G.C.P., 1998, Stress coupling between earthquakes
in northwest Turkey and the north Aegean Sea: Journal of Geophysical Research,
v. 103, p. 24,46924,486.
National Institute of Building Sciences, 1997, HAZUS97Users Manual: Washington,
D.C., unnumbered pages; available from National Institute of Building Sciences,
1201 L St. NW, Suite 400, Washington, D.C., 20005.
National Institute of Standards and Technology, 1996, The January 17,1995 Hyogoken-
Nanbu (Kobe) earthquake: National Institute of Standards and Technology Spe-
cial Report 901, 538 p.

63


64

ORourke, T.D., ed., The Loma Prieta, California, earthquake of October
17, 1989Marina district: U. S. Geological Survey Professional
Paper 1551F, 215 p.
Rockwell, T., Barka, A., Dawson, T., Thorup, K., and Akyuz, S., in press,
Paleoseismology of the Gazikoy-Saros segment of the North Anatolia
fault, northwestern TurkeyComparison of the historical and pale-
oseismic records, implications of regional seismic hazard, and mod-
els of earthquake recurrence: International Journal of Seismology.
Stein, R.S., Barka, A.A., and Dieterich, J.H., 1997, Progressive failure on
the North Anatolian fault since 1939 by earthquake stress triggering:
Geophysical Journal International, v. 128, p. 594604.
Straub, C.S., and Kahle, H., 1995, Active crustal deformation in the Mar-
mara Sea region, NW Anatolia, inferred from GPS measurements:
Geophysical Research Letters, v. 22, no. 18, p. 25332536.
Trifunac, M.D., and Brady, A.G., 1975, A study on the duration of strong
earthquake ground motion: Seismological Society of America Bulle-
tin v. 65, no. 3, p. 581626.
U.S. Geological Survey, 1989, Lessons learned from the Loma Prieta, Cal-
ifornia, earthquake of October 17, 1989: U.S. Geological Survey Cir-
cular 1045, 48 p.
U.S. Geological Survey, 1996, USGS response to an urban earthquake:
U.S. Geological Survey Open-File Report 96263, 78 p.
Wood, H.O., and Neumann, F., 1931, Modified Mercalli intensity scale of
1931: Bulletin of the Seismological Society of America, v. 21, no. 4,
p. 277283.

B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
B
L
E
E
D
BLEED
F
O
L
D
F
O
L
D
Manuscript approved for publication November 22, 1999
Published in the Central Region, Denver, Colorado
Graphics coordinated by Carol A. Quesenberry
Design and photocomposition by Carol A. Quesenberry
Edited by S.J. Kropschot

Ci2010Jul011724424296629 USGS Risk Reduction USGSC 200

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ci2010Jul011724424296629 USGS Risk Reduction USGSC 200

Uploaded by

Copyright:

Available Formats

Implications for

Earthquake Risk Reduction

100 0 200 KILOM ETERS

0 100 200 KILOM ETERS

You might also like